The Relationship Between Increased Police Patrols and Violent Cri PDF

Walden University
ScholarWorks
Walden Dissertations and Doctoral Studies
Walden Dissertations and Doctoral Studies
Collection
2016
The Relationship Between Increased Police Patrols

and Violent Crime Rates in Seven United States
Cities
Victor Sylvester Kabia
Walden University
Follow this and additional works at: http://scholarworks.waldenu.edu/dissertations

Part of the Public Administration Commons, and the Public Policy Commons
This Dissertation is brought to you for free and open access by the Walden Dissertations and Doctoral Studies Collection at ScholarWorks. It has been
accepted for inclusion in Walden Dissertations and Doctoral Studies by an authorized administrator of ScholarWorks. For more information, please
contact ScholarWorks@waldenu.edu.
Walden University
College of Social and Behavioral Sciences
This is to certify that the doctoral dissertation by
Victor Kabia
has been found to be complete and satisfactory in all respects,

and that any and all revisions required by
the review committee have been made.
Review Committee
Dr. Hilda Shepeard, Committee Chairperson,
Public Policy and Administration Faculty
Dr. Mark Stallo, Committee Member,

Dr. Tanya Settles, University Reviewer,

Chief Academic Officer

Eric Riedel, Ph.D.
Walden University
2016
Abstract
The Relationship Between Increased Police Patrols and Violent Crime Rates in Seven
United States Cities
by
MALS, Georgetown University, 2005
BA Hon., Fourah Bay College, University of Sierra Leone, 1981
Dissertation Submitted in Partial Fulfillment
of the Requirements for the Degree of
Doctor of Philosophy
Public Policy and Administration
Walden University
May 2016
Abstract
Large, metropolitan areas across the nation have experienced high rates of violent crime
over the past 2 decades. As a consequence, law enforcement agencies have increased
patrol efforts, but little is known about whether the decrease in violent crime rates was
correlated to increased police patrols or to the economic variables of unemployment,
inflation, level of education, unemployment compensation, and homeownership. The
purpose of this non-experimental, correlational study was to examine the nature of the
relationship between increased police patrols, the 5 economic variables, and violent crime
rates in 7 large US cities for a 10-year period. The theoretical framework for this study
was based on Paternoster’s deterrence theory and Becker’s economic theory of crime
causation. Data were acquired from the Federal Bureau of Investigation and used a
sample of 114 cases of reported violent crimes for each city included in the study for the
years 2000 – 2010 (n = 798). A multiple regression analysis was initially performed with
inconclusive results. Spearman’s correlations between each of the independent and
dependent variables of violent crime indicated that all the independent variables except
for homeownership had statistically significant inverse correlations with violent crime
rates. The findings of this study may be used by law enforcement agencies and policy
makers to develop crime prevention interventions that address those economic factors
associated with violent crime, thereby promoting positive social change through creating
safer communities.
The Relationship Between Increased Police Patrols and Violent Crime Rates in Seven
United States Cities
by
MALS, Georgetown University, 2005
BA Hon., Fourah Bay College, University of Sierra Leone, 1981
Dissertation Submitted in Partial Fulfillment
of the Requirements for the Degree of
Doctor of Philosophy
Public Policy and Administration
Walden University
May 2016
Dedication
I dedicate this dissertation with affection to my father, mother, and all my
teachers and dedicated friends who by their interest and encouragement have continually
inspired me.
Acknowledgments
I am grateful to Dr. Hilda Shepeard, Dr. Stallo, and Dr. Settles, my dissertation
committee for their invaluable guidance and all my tutors at Walden University who in
various ways have brought me to this desired point of my quest.
In the same vein, I wish to thank Alex Naron, my computer mentor, and Carolyn
Suitt, the indomitable cheerleader, who cheered me to the end. To my numerous friends
at the Center for Information Technology (CIT), Andy, Francini, Harry, Bill, Peter,
Aloba, Edward, Agnes, Caponitti, and all my colleagues at MVM and many others for
their encouragement.
To my family, Marie, Hannah, Samuel R. Kabia, Victoria, John, Sentho, Sinnah,
Gbessay, Ya Marie, Bafia, Baby Abu, Potho, Lombeh, Cecilia, Tutu, Fatmata, and
Namina, I say “Thank You” for believing in me.
To my YUSA family, Samuel B. Koroma, Ibrahim Kamara, Sallu Fullah, Musa
Bangura, Sallieu Kamara, Beatrice Conteh, Beatrice Mulwanda, Abdul Kamara, etc.,
thank you for your encouragement.
Finally, I thank all my sons-in- law, Abdul Rahman Kargbo, Idrissa Sesay,
Desmond, and all those related to this family in one way or the other.
Table of Contents
List of Tables .......................................................................................................................v
List of Figures .................................................................................................................... vi
Chapter 1: Introduction ........................................................................................................1
Background ....................................................................................................................1
Problem Statement .........................................................................................................2
Purpose of the Study ......................................................................................................3
Research Questions and Hypotheses .............................................................................4
Theoretical Framework for the Study ............................................................................8
Deterrence Theory .................................................................................................. 9
Social Control Theory ........................................................................................... 10
Economic Theory of Crime Causation ................................................................. 12
Nature of the Study ......................................................................................................14
Definition of Terms......................................................................................................15
Assumptions.................................................................................................................15
Scope and Delimitations ..............................................................................................16
Limitations ...................................................................................................................16
Significance of the Study .............................................................................................17
Summary ......................................................................................................................19
Chapter 2: Literature Review .............................................................................................23
Introduction ..................................................................................................................23
i
Literature Search Strategy............................................................................................24
Theoretical Foundation ................................................................................................25
Deterrence Theory and Crime ............................................................................... 25
Social Control Theory and Crime ......................................................................... 27
Economic Theory of Crime Causation ................................................................. 28
Literature Review Related to Key Variables and Concepts.........................................30
Ineffectiveness of Police Patrols as Deterrence to Crime ..................................... 30
Research Establishing the Effectiveness of Police Patrols ................................... 35
Limited value of traditional patrol methods: Routine and random ................. 36
Effectiveness of directed police patrols .......................................................... 37
Role of foot patrol in crime reduction............................................................. 42
Police Numbers do Matter in Crime Reduction .................................................... 45
Hot-Spot Policing.................................................................................................. 49
Addressing the Endogeneity Problem ................................................................... 54
Other Factors That Affect the Rates of Violent Crime ................................................66
Unemployment ...................................................................................................... 66
Education .............................................................................................................. 67
Homeownership .................................................................................................... 68
Cost of Living (Inflation) ...................................................................................... 68
Unemployment Compensation.............................................................................. 69
Summary and Conclusion ............................................................................................70
ii
Chapter 3: Methodology ....................................................................................................75
Introduction ..................................................................................................................75
Research Design and Rationale ...................................................................................75
Methodology ................................................................................................................78
Population ............................................................................................................. 78
Sampling and sampling procedures ................................................................ 79
Sampling plan ................................................................................................. 80
Data Collection ..................................................................................................... 80
Operationalization of Constructs .................................................................................82
Data Analysis Plan ................................................................................................ 82
Research Questions ............................................................................................... 83
Study Variables ............................................................................................................87
Dependent Variable .............................................................................................. 87
Independent Variables .......................................................................................... 87
Unemployment ................................................................................................ 88
Years of education .......................................................................................... 88
Homeownership .............................................................................................. 88
Cost of living (inflation rates) ......................................................................... 88
Unemployment compensation ........................................................................ 88
Preliminary Analysis ....................................................................................................88
Ethical Considerations .................................................................................................92
iii
Threats to Validity .......................................................................................................93
Summary ......................................................................................................................93
Chapter 4: Results ..............................................................................................................95
Description of the Sample............................................................................................95
Detailed Analysis .......................................................................................................103
Normality Assumption ........................................................................................ 104
Linearity Assumption.......................................................................................... 105
Homoscedasticity Assumption............................................................................ 111
Absence of Multicollinearity Assumption .......................................................... 112
Spearman Correlations ........................................................................................ 114
Summary ....................................................................................................................116
Chapter 5: Discussion, Conclusions, and Recommendations ..........................................118
Summary of the Findings ...........................................................................................118
Interpretation of Findings ..........................................................................................119
Limitations of the Study.............................................................................................121
Recommendations for Future Studies ........................................................................123
Conclusions ................................................................................................................124
References ........................................................................................................................125
iv
List of Tables
Table 1. Descriptive Statistics of Continuous Variables .................................................. 96
Table 2. Multiple Linear Regression with Six City Characteristics Predicting Violent
Crime Rates ............................................................................................................. 114
Table 3. Spearman Correlations Between Six City Characteristics and Violent Crime
Rates ........................................................................................................................ 116
v
List of Figures
Figure 1. The effects on local police from changes in state tax rates. .............................. 59
Figure 2. Bar chart for violent crime rates by year (2000–2010). .................................... 97
Figure 3. Bar chart for number of police patrols by year (2000–2010). ........................... 98
Figure 4. Bar chart for unemployment rate by year (2000–2010). ................................... 99
Figure 5. Bar chart for average years of education by year (2000–2010). ..................... 100
Figure 6. Bar chart for home ownership rates by year (2000–2010). ............................. 101
Figure 7. Bar chart for cost of living index by year (2000–2010). ................................. 102
Figure 8. Bar chart for average weekly unemployment compensation by year (2000–
2010). ...................................................................................................................... 103
Figure 9. Normal P-P scatterplot to assess normality for city characteristics and violent
crime rate. ............................................................................................................... 104
Figure 10. Scatterplot to assess for linearity between number of police patrols and violent
crime rate. ............................................................................................................... 106
Figure 11. Scatterplot to assess for linearity between unemployment rate and violent
crime rate. ............................................................................................................... 107
Figure 12. Scatterplot to assess for linearity between average years of education and
violent crime rate. ................................................................................................... 108
Figure 13. Scatterplot to assess for linearity between home ownership rate and violent
crime rates. .............................................................................................................. 109
vi
Figure 14. Scatterplot to assess for linearity between cost of living index and violent
crime rate. ............................................................................................................... 110
Figure 15. Scatterplot to assess for linearity between average weekly unemployment
compensation and violent crime rate. ..................................................................... 111
Figure 16. Residuals scatterplot for homoscedasticity for city characteristics predicting
violent crime rates. .................................................................................................. 112
vii
1
Chapter 1: Introduction
The plague of violent crime in Washington, DC; Indianapolis, IN; Philadelphia,
PA; Boston, MA; Kansas City, MO; Los Angeles, CA; and Pittsburgh, PA has caused a
large deal of human misery as people have lost loved ones in the spate of violence.
Systematic, scientifically-based research focuses efforts to reduce uncertainty and
increase efficacy regarding problems, such as the rate of violent crime (Knoke,
Bohrnstedt, & Mee, 2002).
Background
One of the defining features of American democracy is the opportunity for free
and open debate on critical public policies, such as the best strategy for reducing or
preventing violent crime (Akesson et al., 2012). The debate regarding the prevention of
crime, including violent crime, is based on the ideologies of the major political parties,
Democrats and Republicans, or liberals and conservatives, respectively. The existence of
a strong social support system is one of the cornerstones of liberal politics, whereas
conservative ideology relies on the widespread institution of police agents and deterrence
(Akesson et al., 2012).
Akesson et al. (2012) asserted that conservatives are critical of liberals for the
failed policies in controlling crimes. Wilson (1975) openly criticized "the liberal
bankruptcy" of the liberal social support policies, including the liberals' policies in the
1960s and 1970s that failed to decrease or prevent crime in the United States, and
concluded that their policies had no relationship with crime (Ren, L., Zhao, J., and
2
Lovrich, N. P. 1990 p.316). The liberals fought back 10 years later and retorted that
conservative policies had not been successful in reducing or preventing crime either
(Currie, 1985). This study helped to fill the gap in the literature regarding how the
outcomes of existing empirical studies could be related to other cities not included in
previous meta-analyses (Gill, Weisburd, Telep, Vitter, & Bennett, 2014; Hoelzer &
Gorman, 2011; Philadelphia Foot Patrol Experiment, 2009). The integrative nature of the
quantitative research study with the stated independent variables may provide a more
comprehensive understanding of the relationship between these variables and the rate of
violent crime. An understanding of the relationship between the independent and the
dependent variables will help policy makers in the wise allocation of resources to fight
violent crime.
Problem Statement
Washington, DC; Indianapolis, IN; Philadelphia, PA; Boston, MA; Kansas City,
MO; Los Angeles, CA; and Pittsburgh, PA have experienced high levels of violent crime
for the past 2 decades. Even when violent crime fell in some of the aforementioned cities
across the United States, those cities still had high levels of violent crime (Gill et al.,
2014). Some researchers asserted that increases in police patrols had nothing to do with
the reduction of violent crime, but rather attributed the change in violent crime rates to
other factors, such as unemployment (Rafael & Winter-Ember, 2000), cost of living
(inflation rates; Seals & Nunley, 2007), years of education (Lochner & Moretti, 2001),
homeownership (Ni & Decker, 2009), and unemployment compensation (Levitt, 2001). I
3
analyzed the nature of the relationship between each of the independent variables (police
patrols, unemployment, cost of living [inflation], years of education, homeownership, and
unemployment compensation) and the dependent variable (the rate of violent crime).
For policy makers to design strategies that reduce the rate of violent crime, data to
support effective strategies so that resources are not wasted must exist. Through the
study, I determined whether increased police patrols and the other independent variables
in the seven cities are related to violent crime rates in each of these cities. The strength
and direction of each relationship, if statistically significant, could assist policy makers to
make better choices in resource allocation.
Purpose of the Study
The purpose of this study was to determine the nature of the relationship between
the independent variables (police patrols, unemployment rate, years of education, home-
ownership, cost of living [inflation], unemployment compensation), and the dependent
variable (the rate of violent crime) in seven major U.S. cities. The hypothesis was tested
using both correlational and regression analyses. Absolute causality was not determined;
the study involved the method of quantitative numerical assessment and statistical
procedures using a nonexperimental correlational regression research design to test the
hypotheses and to evaluate the nature of the relationships between the independent
variables and dependent variable. This study’s contents contributed to the current state of
knowledge regarding the historical (2000–2010) relationship between the stated

4
independent variables and the dependent variable of the rate of violent crime in the seven
selected cities with high rates of violent crime.
Research Questions and Hypotheses
A debate regarding the role and relationship of the police or police patrols with
the rate of violent crime among lawmakers, criminologists, researchers, and law
enforcement agencies exists (Gill et al., 2014; Hoelzer & Gorman, 2011; Ritchey &
Nicholson-Crotty, 2011; Vollaard, 2005; Vollaard & Hamed 2009). Some criminologists
have held the position that the police are not critical to the solution of crime and that no
relationship exists between the police and reduction in crime (Marvell & Moody, 1996).
Others have asserted that the police could be critical in crime prevention or reduction
(Vollaard, 2005). With this debate in mind, I developed the following research and
hypotheses to guide my study:
Research Question 1: What is the nature of the relationship between the rate of
police patrols (as measured by foot patrols, automobile patrols, and motorcycle
patrols) and the rate of violent crimes per 100,000 residents in the seven cities
examined in aggregate during the period 2000–2010?
Null Hypothesis 1: There is no statistically significant relationship
between the rate of police patrols (as measured by foot patrols, automobile
patrols, and bicycle patrols) and the rate of violent crimes per 100,000
residents in the seven cities examined in aggregate during the period
2000–2010.
5
Alternate Hypothesis 1: There is a statistically significant relationship
2000–2010.
unemployment (as measured by the monthly unemployment rate of each city) and
the rate of violent crimes per 100,000 residents in the seven cities examined in
aggregate during the period 2000–2010?
between the rate of unemployment (as measured by the monthly
unemployment rate of each city) and the rate of violent crimes per 100,000
2000–2010.
2000–2010.
Research Question 3: Based on archival data (2000–2010), what is the
relationship the between the level of education (as measured by the average
6
number of months a citizen of the city has completed school) and the rate of
violent crimes per 100,000 residents in the seven cities examined in aggregate?
between the level of education (as measured by the average number of
months a citizen of the city has completed school) and the rate of violent
crimes per 100,000 residents in the seven cities examined in aggregate
during the period 2000–2010.
Research Question 4: What is the relationship between the rate of home
ownership (as measured by the monthly home ownership rates of each city) and
between the rate of home ownership (as measured by the monthly home
ownership rates of each city) and the rate of violent crimes per 100,000
2000–2010.
7
2000–2010.
Research Question 5: What is the relationship between the cost of living (as
measured by annual inflation rates) and the rate of violent crimes per 100,000
residents in the seven cities examined in aggregate during the period 2000–2010?
between the cost of living (as measured by annual inflation rates) and the
rate of violent crimes per 100,000 residents in the seven cities examined in
aggregate during the period 2000–2010.
Research Question 6: What is the relationship between the rate of unemployment
compensation (as measured by the weekly unemployment compensation) and the

8
between the rate of unemployment compensation (as measured by the
weekly unemployment compensation) and the rate of violent crimes per
100,000 residents in the seven cities examined in aggregate during the
period 2000–2010.
period 2000–2010.
Theoretical Framework for the Study
The framework that grounded this study was the concept of crime deterrence or
reduction, originating from the writings of Beccaria (1738–1794) in Italy and Bentham
(1748–1832) in England. Deterrence theory grounded this study to a larger theoretical
framework because of the emphasis on deterrence via punishment and sanctions to
prevent or minimize criminal behaviors in individuals and the population at large
(Ritchey & Nicholson-Crotty, 2011). Because this study pertained to the relationship
between the six independent variables and the dependent variable, deterrence theory,
supported by social control theory and economic theory of crime causation, comprised an
appropriate framework for this study.

9
Deterrence Theory
The focus of deterrence theory is that punishment serves as a way to minimize
and prevent criminal behaviors (Ritchey & Nicholson-Crotty, 2011). Deterrence operates
in two ways: (a) specific and (b) general (Paternoster, 2010). Specific deterrence refers
to the punishment of an individual to discourage him or her from committing another
crime, whereas general deterrence refers to punishment given to an individual with the
intent of making an example for the general population.
Empirical support for deterrence theory is mixed (Piquero, Paternoster, Pogarsky,
& Loughran, 2011). While some studies show support for deterrence theory, providing
empirical evidence that sanctions can prevent or minimize offenses (Pauwels, Weerman,
Bruinsma, & Bernasco, 2011), other studies show that deterrence theory is more effective
in deterring some types of crime than others (Durlauf & Nagin, 2011; Tonry, 2008).
Pauwels et al. (2011) showed that sanctions can decrease adolescent offending and that
morality may also have an influence on the relationship of crime and punishment,
suggesting that “the less a person morally supports specific types of offending, the more
strongly that person is affected by perceived sanctions” (Pauwels et al., 2011, p. 386).
Tonry (2008) found mixed results in establishing punishment as a deterrent factor
in offending. In more serious crimes, Tonry suggested that sanctions are not effective
deterrents for specific or general purposes. Sanctions may be more effective in
nonserious crimes, such as traffic violations, tax evasion, and speeding offenses.
10
In a literature review conducted by Durlauf and Nagin (2011), the researchers
stated that lengthy prison sentences do not deter criminal offenses, suggesting that long
sentences may not be an effective form of punishment. Durlauf and Nagin stated that
sanctions might be effective deterrents based on the perceived risk and severity of the
sanction. Social control theorists believe that any person would commit a crime if that
person could go undetected (Ritchey & Nicholson-Crotty, 2011). Many U.S. cities have
utilized deterrence theory and increased police presence with success.
Social Control Theory
The social control theory relating to social support is an alternative view on crime,
placing emphasis on strengthening the relationships of individuals in order to avoid
deviant and antisocial behaviors instead of relying on punishment (Hirschi, 2001). The
assumption of social control theory is that an individual will violate the law if he or she
can get away with it (Ritchey & Nicholson-Crotty, 2011). Family relationship is one
social relationship that has significance to this theory because of the role of family in the
lives of individuals. Compared to deterrence theory, social control theory relating to
social support has a more positive view on criminals, emphasizing positive reinforcement
instead of punishment.
Researchers have conceptualized social control in various forms, which include
attachment, self-esteem, school climate, and family systems (Chen, Cheng, Liang, &
Sato, 2012; Dymnicki, 2014; Peguero, Popp, Latimore, Shekarkhar, & Koo, 2010;
Rutherford, 2011). The finding of most studies pertaining to social control theory is that
11
positive social bonds are negative predictors of delinquency and criminal behaviors
(Boman, Krohn, Gibson, & Stogner, 2012; Chen et al., 2012). Positive social bonds can
be manifested in the relationship of individuals to family, friends, school, and the
community (Chen et al., 2012; Rutherford, 2011).
Most researchers have found that positive parenting prevents criminal and
antisocial behaviors (Dymnicki, 2014). For instance, parent-child problems and
maltreatment can be a predictor of future delinquency. However, Chen et al. (2012) did
not find any significant relationship between the two variables.
In a path analysis study conducted by Rutherford (2011), the researcher combined
social control theory and differential association theory to explain delinquency in terms
of familial variables. The researcher found that positive self-esteem is an antecedent of
juvenile delinquency (Rutherford, 2011). Rutherford also found that relationship with
peers who are engaged in delinquent activities and familial stressors are predictive of
juvenile delinquency. Delinquents tend to have a stronger or more intense type of
friendship compared to nondelinquents (Boman et al., 2012).
School environment was the focus of Chen et al.’s (2012) study on delinquency
using the social control theoretical framework. In terms of the general relationship of
school and delinquency, school climate was a predictor of risky behaviors and
delinquency. With regard to specific school climate factors, Chen et al. found that
children who are engaged in positive prosocial activities, such as sports and church
activities, are less likely to engage in delinquent activities.

12
Although social support, a variation of social control, seems to be an alternative
strategy to deterrence in crime reduction, this study pertained to investigating if a
relationship exists between violent crime and the independent variables of police patrols,
unemployment, homeownership, years of education, cost of living (inflation), and
unemployment compensation. I did not focus on finding out which of the two strategies,
social support and deterrence, is better in crime reduction.
Economic Theory of Crime Causation
This study was also grounded in the economic theory of crime causation.
Researchers have mostly focused on economic analysis of crime regarding the effects that
incentives have on criminal behavior and the evaluation of alternative methods to bring
about the reduction of crime (Kleemans, 2012). Economic theories on crime have proven
the causes of crime to be diverse, and Becker (1968) confirmed the independent variables
of unemployment, lack of education (or marketable skills), and cost of living (inflation)
to be related to crime rates. Economic theories of crime also contend that an alternative
to the deterrence theory would be to improve on the factors that drive people into
criminal activities. For example, if schools receive no funding and are therefore
ineffectual, people cannot continue on to higher education or get high-paying jobs,
resulting in a lack of financial resources. This study involved addressing a decline in
wages, inflation or high cost of living, and unemployment as alternatives to sanctions and
punishment in fighting crime. The economic theory asserts that criminals make choices
after a cost-benefit analysis regarding their engagement in criminal activities. If people

13
perceive that legitimate employment earns them low wages compared to illegitimate
activities, they might take the risk of being caught and punished (Kleemans, 2012).
Economic theory asserts that a relationship exists between the dependent variable
of crime and the independent variables of unemployment, years of education,
homeownership, cost of living (inflation), and unemployment compensation (Becker,
1968). Through this study, I determined the nature of the relationship between these
variables, police patrols, and violent crime rates. Determination of the strength and
direction of the relationship between these independent variables and the dependent
variable of violent crime rates occurred using correlation and multilinear regression. The
value of this study was also regarding the integrative approach, using both deterrence
theory and the economic theory strategy. The correlation and multilinear regression
analyses revealed the nature of the relationship between these independent variables and
the dependent variable. The strength and direction of these relationships may be useful
guides to policy makers and lawmakers in decision-making and prediction of violent
crime (Kruger, 2011).
In the 18th century, legal and penal reform addressed deterrence theory or
doctrine. The main features of the deterrence theory of crime by Beccaria and Bentham
were certainty, severity, and celerity of punishment (Ritchey & Nicholson-Crotty, 2011).
I decided to use the theory of deterrence in this study and not the theory of desistance
because the study was about crime prevention and not about why criminal offenders stops
14
committing crime, or desistance. In addition, desistance had definitional, measurement,
and theoretical incoherence with study’s focus (Paternoster, 2010).
The role of the police and police patrols became prominent in deterrence theory
in the late 20th and early 21st centuries with the development of control theories, such as
social control theory. When the World Health Organization asked why many people
conformed with legal and cultural norms, control theorists believed that social controls
prevented people from committing crimes and that crimes would be committed if the
social controls were broken (Lauritsen, Rezey, & Heimer, 2014). Social controls seem to
have an adverse relationship with the development of implicit morality (Hirschi, 2001).
Nature of the Study
This study followed a research design involving correlational archival data from
2000–2010, collected from each of the seven cities with regard to the rate of violent
crimes. The data used were open sourced data retrieved from Federal Bureau of
Investigation (FBI) databases (see Appendix X). A quantitative approach is ideal for
examining the nature of the relationships between the independent variables and the
dependent variable (Neuman, 2005).
Creswell (2009) asserted that a correlation design examines the relationship
between two or more variables in a single group and further stated that a regression can
also predict a relationship between independent variables and the dependent variable.
Kruger (2011) pointed that although a relationship may exist between variables in a
correlation, researchers should not conclude that one independent variable is causing the
15
dependent variable to change. Causation may be a result of the effect of two or more
independent variables on the dependent variable (Creswell, 2009). I therefore determined
the strength and direction of the relationship between each of the six independent
variables and dependent variable.
Definition of Terms
Police Patrol: A police patrol is a police officer or a group of police officers who
have the responsibility to enforce laws that could include traffic control, crime
prevention, and control of crowds at a given area, which can include on foot, in a vehicle,
or on a motorcycle (O'Connor, 2010).
Assumptions
I held the following assumptions regarding the study:
1. The historical data were accurately entered into the retrieval database.
2. The study variables were available in a coherent, analyzable form.
3. All unmeasured confounding variables were accounted for in the
interpretation of the findings.
4. The sample studies analyzed in the study had external validity considerations
that could be used for generalization (Trochim, 2001).
5. The data reported were reliable and met the states’ and FBI’s quality
guidelines.
16
Scope and Delimitations
The scope of this study was limited to making conclusions regarding the
relationship of six independent variables with the dependent variable, both individually
(correlations) and as a group (linear regression). Predictions were only made from the
results of the regression analysis; no absolute causal relationships were made. The only
independent variables examined were the six listed in this chapter to assess their potential
as predictors of rates for violent crime in an aggregate group of seven cities. The study
did not include controlling for extraneous variables so that results produced absolute
causal conclusions. To assert absolute causation, one needs to take into consideration the
research design (Kruger, 2011).
The study did not involve addressing the problem of crime displacement—the
movement of crime from one place to another as a result of a large number police in the
original place—apart from the analysis of some of the studies included in the literature
review (Levitt, 2004). The sample only included 114 cases of violent crimes in each city
(N > 798) during the period 2000–2010, based on the power analysis calculations further
detailed in Chapter 3 and included in Appendix X.
Limitations
I carefully selected historical empirical research studies on the basis of the rigor
of quantitative correlational research designs using archival data. While quantitative
studies can be useful for statistically analyzing research questions and hypotheses, such
designs are unable to measure adequately the depth of experiences inherent within a
17
qualitative study. In addition, some confounding variables may account for the
relationships established in the findings. I acknowledged this limitation, and interpreted
statistical findings with caution. Finally, the selected cities were included based on the
literature indicating the presence of significant trends on rates of violent crime that have
withstood the processes of validity and reliability defense.
Significance of the Study
Various prior researchers (Gill et al., 2014; Hoelzer & Gorman, 2011;
Philadelphia Foot Patrol Experiment, 2009) conducted studies in many cities in the
United States and other cities in Western Europe. These researchers have asserted that
increased police patrols on the streets of a city or town can deter or reduce violent crime
(Draca, Machin, & Witt, 2011; Duborg, Hamed, & Thomas, 2005; Vollaard & Hamed,
2005). However, some of these researchers did not use an integrative approach that
combined the economic variables and the increased police patrol variable to determine
how each variable was related to violent crime and to determine the findings' relevance to
other cities.
The findings of this study could also be applicable to other cities in the United
States. The value of this study existed in providing reliably obtained data and validly
constructed findings that policy makers or academicians can use to further a useful
conversation about determining a strategy to fight violent crime effectively. I hoped to
establish an understanding of the nature of the relationship between the independent
variables and the rate of violent crime by analyzing the contribution that several
18
independent variables make to the relative change in a dependent variable (Kruger,
2011). By using correlation analyses, I also intended to give a mathematical
measurement of the statistical significance and strength of the relationship between the
six independent variables and the dependent variable of the rate of violent crime.
Contrary to the view that correlational studies are of little value to social science
research because they fall short of establishing absolute causation, researchers usually
conduct these studies to address the questions of whether variables are significantly
related or not and to what degree and direction the variables are related (Kruger, 2011).
When a relationship is established in terms of strength and direction in a correlational
study, it forms the basis for prediction, which is usually done before making expenditure
decisions. The use of regression analysis is beneficial for public policy decisions
(Bernard, 2000). Multiple regressions also help to develop theories of causation that
have roots in the relationships between strength and direction (Kruger,2011).
As Vollaard (2006) stated, the two most important functions of government are to
fight crime and to maintain peace and order; however, conflicting information exists
regarding the methods that need to be taken. In addition to the conflicting information
that policy makers have about the best strategy to fight crime, policy makers also face a
scarcity of resources because of the multifarious problems affecting communities. It
would be in the interest of taxpayers that the money spent by government accomplished
something. Taxpayers’ money can be wisely spent if the policy makers have empirical
19
information to help make their decisions related to expenditures on the independent
variables and the rate of violent crime.
If policy makers reach a decision that more police patrols would be the answer in
combating violent crime, it would involve hiring more police officers (Tabarrok, 2004).
Hiring police in most jurisdictions has been and is still a perennial budget management
issue (Groff, Johnson, Ratcliffe, & Wood, 2013). If the decrease of violent crime is more
empirically related to one or more of the economic variables rather than police patrols,
lawmakers should also base their decisions on those findings.
This study could bring about social change because policy makers may use data
provided in this research to guide their decisions about effective strategies to reduce
violent crime. If the use of data enables policy makers to make informed decisions in
fighting violent crime, it could bring about positive social change.
Summary
Criminologists, social scientists, and many other researchers still debate the
efficacy of the rate of police patrols in relation to the rate of violent crime. Conservatism
and liberalism are at the opposite extremes of the continuum regarding the strategies used
in solving and deterring crimes in the United States (Akesson et al., 2012). One
proposition of using the police to reduce or deter violent crime was countered by a
prolific corpus of research. The studies discussed in the literature review in Chapter 2
expound on this scholarly argument. Marvell and Moody (1996) found, in a more
rigorous research design, that the rate of police patrols did not negatively correlate with
20
the rate of violent crime. In addition, Marvell and Moody (1996), Gill et al. (2014), and
Kovandzic and Sloan (2002) found strong evidence that the rate of police patrols was
significantly and negatively related to rates of violent crime, especially homicide rates at
the city and state levels of analysis. Chapter 3 presents further information on these rates.
A shift has taken place in support of the theory that a positive and statistically
significant correlation exists between the rates of police patrol and rates of violent crime
because of the solution that many researchers have found for the problem of endogeneity.
Research methods and tools have improved with more powerful and more accurate
software for measurement. In the late 20th and early 21st centuries, most of the research
showed that increased police patrols decreased violent crime (Vollaard & Hamed, 2009).
Many researchers found that the methodology used by prior researchers was problematic.
The statistical analyses used by prior researchers were not suitable to solve the problems
of simultaneity and the incomplete control variables biases (Kovandzic & Sloan, 2002).
Through this study, I examined the efficacy of the deterrence theory, as
originating from the works of Bentham and Cesare (Paternoster,2010). The theory
operates under the framework that increased police or increased police patrols will reduce
violent crimes (Ritchey& Nicholson-Crotty, 2011). This study involved analyzing prior
studies conducted in many cities in the United States to examine whether increased police
patrols can deter or reduce violent crime in Washington, DC; Indianapolis, IN;
Philadelphia, PA; Boston, MA; Kansas City, MO; Los Angeles, CA; and Pittsburgh, PA.
Braga (2006), Braga et al. (2001), Cohen and Ludwig (2003), Draca et al. (2008), and
21
Hoelzer and Gorman (2011) have shown that increased police patrols are correlated with
reduced violent crime.
The results of this study may have implications for social change because
knowing what works to deter violent crime could help lawmakers make decisions. Many
cities, including Washington, DC, spend a lot of money and resources to fight violent
crime. For instance, the Justice Department awarded a $1 million grant to enable the
Washington, DC Metropolitan Police Firearms Identification Unit to fight against violent
crime (National Association to Restore Pride in America’s Capital [NARPAC], 2007). In
2006, when violent crime increased in DC, police patrol overtime increased to $4.2
million (NARPAC, 2006). The cost of fighting violent crime has put responsibility on
lawmakers to know what strategies work to reduce violent crime and thereby avoid
wasteful spending. Policy makers need to stop and redirect resources expended on
ineffective strategies to effective strategies. Cities with high crime rates usually suffer
from low economic growth. In addition, cities or towns with low crime rates are more
likely to attract investors, thereby improving the socioeconomic status of their
communities (Kulawczuk, 1998). Therefore, lawmakers might find it useful to know if
more police patrols could deter violent crime to enable themselves to make educated
decisions for the benefit of their communities.
Researchers studied extensively in the 1970s, 1980s, and 1990s to address this
public policy issue but did not arrive at a conclusion. The researchers in the 1980s and
1990s found that police patrols on the streets of towns or cities negatively correlated with
22
the rate of violent crime. Chapter 2 details these findings in the literature review and
highlights the progression of the research from the 1970s to the present. In the literature
review, I examined, analyzed, and synthesized the past and current literature on the
subject to establish the foundation for better understanding of the research. Chapter 3
outlines the methodology of the study.

23
Chapter 2: Literature Review
Introduction
Washington, DC; Indianapolis, IN; Philadelphia, PA; Boston, MA; Kansas City,
MO; Los Angeles, CA; and Pittsburgh, PA have experienced high levels of violent crime
for the past 2 decades. Even when violent crime fell in some of the aforementioned cities
across the United States, those cities still had high levels of violent crime (Gill et al.,
2014). This study analyzed the nature of the relationship between each of the
independent variables (police patrols, unemployment, cost of living [inflation], years of
education, homeownership, and unemployment compensation) and the dependent
variable (the rate of violent crime). The data used in the study came from the Uniform
Crime Reporting Program that has been vetted by Federal Bureau of Investigation (FBI).
The FBI Uniform Crime Reporting (UCR) Program collected the violent crime
data reported for the seven cities to ensure that the data collection followed the guidelines
(https://www.fbi.gov/about-us/cjis/ucr/data_quality_guidelines). Submitting UCR data to
the FBI is a collective venture on the part of the city, county, and state (p.1). The state
UCR program reports crimes using the uniform offense definitions to a centralized
repository within their state (p.1). The state depository forwards the data to the FBI after
clearing the reports of any inaccuracies (p.2). Those in states who do not have a state
UCR program send their data directly to the FBI (p.3). Staff members at the FBI review
the information or data for accuracy and reasonableness and enter the data into the
national database (p.3).

24
The FBI has some basic procedures to ensure that UCR data are accurate, which
are:
1. Check data for its logical consistency.
2. Check data for reasonableness.
3. Ensure that other statistical edit functions are processed within established
parameters (p.3).
Apart from the internal quality assurance measures, the FBI also gives on-site reviews of
the records of the law enforcement agencies that submit UCR data to the FBI (p.4). The
FBI has a team of auditors who conduct periodic reviews of UCR crime data collected
and reported by the program’s data contributors (p.5). Each state UCR Program is
subject to review at least once every 3 years to evaluate the state’s compliance with
national UCR guidelines (https://www.fbi.gov/about-us/cjis/ucr/data_quality_guidelines).
This chapter includes the literature search strategy, theoretical foundation, literature
review related to key variables and concepts, and closes with the summary and
conclusions.
Literature Search Strategy
The research process of peer-reviewed articles began with searches using the
following databases: ABI/INFORM Complete, EBSCOHost, Emerald Management
Journals, SAGE Premier, ProQuest Central, Thoreau, and Google Scholar. I accessed the
aforementioned databases through the Walden University library. I used different
combinations of keywords related to the sections of the literature review to locate peer-
25
reviewed research studies. Some of these keywords included: violent crime, saturated
patrols, routine patrols, random patrols, deterrence theory, economic theory of crime,
causation, correlation, etc. The majority of these articles, from peer-reviewed journals,
were published between 2010 and 2015. The review of professional and academic
literature includes four main sections, with the first topic covering framework for this
study and the three other topics directly relating to the role of police patrols on the
reduction of violent crime.
Theoretical Foundation
Deterrence Theory and Crime
Schelling developed deterrence theory in 1966 as a preventative military strategy
during the Cold War (Paternoster, 2011). The emphasis of deterrence theory is that
punishment serves as a way to minimize and prevent criminal behaviors (Ritchey &
Nicholson-Crotty, 2011; Rizzolli & Stanca, 2012). Deterrence operates in two ways, (a)
specific and (b) general (Paternoster, 2010; Piquero, Paternoster, Pograsky, & Loughran,
2011). Specific deterrence pertains to deterring an individual from committing another
crime, whereas general deterrence pertains to punishing an individual as an example to
the general population (Ritchey & Nicholson-Crotty, 2011).
Empirical support for deterrence theory is mixed (Piquero et al., 2011). Some
studies show support for deterrence theory, providing empirical evidence that sanctions
can prevent or minimize offenses (Pauwels, Weerman, Bruinsma, & Bernasco, 2011).
26
Some studies show that deterrence theory is not effective in deterring all types of crimes
(Durlauf & Nagin, 2011; Tonry, 2011).
Pauwels et al. (2011) examined if sanctions prevent adolescent offending in a
European sample and showed that they could do so. The results also showed that
morality may have an influence on the relationship of sanction and offending, suggesting
that “the less a person morally supports specific types of offending, the more strongly
that person is affected by perceived sanctions” (Pauwels et al., 2011, p. 386). Tonry
(2011) found mixed results in establishing punishment as a deterrent factor. For more
serious crimes, Tonry suggested that sanctions are not effective deterrents. Sanctions
may be more effective in non-serious crimes, such as traffic violations, tax evasion, and
speeding offenses (Tonry). In a literature review conducted by Durlauf and Nagin
(2011), the researchers contended that lengthy prison sentences do not deter criminals,
suggesting that they may not be an effective form of punishment. Durlauf and Nagin
stated that sanctions might be an effective form of deterrent based on the perceived risk
and severity of the sanction.
Past, related studies showed that sanctions can decrease adolescent offending.
Social control theorists assumed that everyone would violate the law if they were not
detected (Ritchey & Nicholson-Crotty, 2011; Rizzolli & Stanca, 2012). Many U.S. cities
have utilized deterrence theory and increased police presence, and were successful
(Ritchey & Nicholson- Crotty). Deterrence theory may contribute to explaining the
results of the study.

27
Social Control Theory and Crime
Social control theory has evolved over time and was adopted and revised to
pertain to crime by Hirschi in 1990 (Papachritos, Meares, & Fagan, 2012). The social
control theory relating to social support is an alternative view on crime, putting emphasis
on strengthening the relationships of individuals to avoid deviant and antisocial behaviors
instead of relying on punishment (Hirschi, 2001). The assumption of social control
theory is that individuals would violate the law if they could get away with it
(Papachritos, Meares, & Fagan, 2012). Family relationship is one social relationship that
has significance regarding this theory because of the role of family in the lives of
individuals. Compared to deterrence theory, social control theory relating to social
support has a more positive view on punishment, emphasizing positive reinforcement
instead of punishment.
Researchers have conceptualized social control in various forms, which include
attachment, self-esteem, school climate, and family systems (Chabay, 2013; Peguero,
Popp, Latimore, Shekarkhar, & Koo, 2010; Pietromonaco, Uchino, & Dunkel Schetter,
2013). The general finding of most studies pertaining to social control theory is that
positive social bonds are negative predictors of delinquency and criminal behaviors
(Boman, Krohn, Gibson, & Stogner, 2012; Hay, Meldrum, & Piquero, 2012). Positive
social bonds can be manifested in the relationship of individuals to family, friends,
school, and the community (Cusick, Havlicek, & Courtney, 2012; Hay et al., 2012).
28
The prevailing finding in terms of the relationship between parents and
delinquency is that positive parenting prevents criminal and antisocial behaviors
(Dymnicki, 2014; Lösel, Stemmler, & Bender, 2013). For instance, parent-child
problems and maltreatment can be a predictor of future delinquency. However, Gault-
Sherman (2012) did not find any significant relationship between the two variables.
In a path analysis study, Rutherford (2011) combined social control theory and
differential association theory to explain delinquency in terms of familial variables. The
researcher found that positive self-esteem is an antecedent of juvenile delinquency
(Rutherford, 2011). The researcher also found that relationships with peers who are
engaged in delinquent activities and familial stressors are predictive of juvenile
delinquency. Delinquents tend to have a stronger or more intense type of friendship
compared to non-delinquents (Boman et al., 2012).
School environment was the focus of Klein, Cornell, and Konold’s (2012) study
on delinquency using the social control theoretical framework. In terms of the general
relationship of school and delinquency, school climate was a predictor of risky behaviors
and delinquency (Boman & et al., 2012). With regard to specific school climate factors,
Chen et al. (2012) found that children who are engaged in positive, prosocial activities,
such as sports and church activities, are less likely to engage in delinquent activities.
Economic Theory of Crime Causation
Another theory on which this study was grounded is the economic theory of crime
causation. Becker (1968) reformed the economic theory of crime causation in 1968 in an
29
effort to assess criminal justice policies. Economic analysis of crime has mostly focused
on the effects that incentives have on criminal behavior and the evaluation of alternative
methods to bring about the reduction of crime (Draca & Machin, 2015). Economic
theories showed the diverse causes of crime, and Becker (1968) advanced independent
variables, such as unemployment, lack of education (or marketable skills), and cost of
living (inflation), to be related to crime rates. Economic theories of crime also contend
that an alternative to the deterrence theory in crime reduction would be to improve on the
factors that drive people into criminal activities (Draca & Machin, 2015). Neglecting to
provide proper education could increase the prevalence of crime. A decline in wages,
inflation or high cost of living, and unemployment were addressed as alternatives to
sanctions and punishment in fighting crime (Becker,1968). The economic theory
asserted that criminals make choices after a cost-benefit analysis regarding their
engagement in criminal activities (Becker). If people perceive that legitimate
employment earns them low wages compared to illegitimate activities, they would take
the risk of being caught and punished (Draca & Machin, 2015).
In a quantitative study, Fagan (2010) explored the deterrence of crime among
juvenile offenders and found that racial disparities and underprivileged economic
conditions directly correlated to an increase in the commission of violent crime among
these individuals. Fagan also asserted that these individuals often committed crimes as a
reaction to their economic surroundings. Furthermore, individuals who became repeat
offenders did so in response to the actions of their peers. Fagan further contended that
30
the best response to reducing these offenses would be to focus on deterring the
opportunity for crime in their environment.
Literature Review Related to Key Variables and Concepts
Ineffectiveness of Police Patrols as Deterrence to Crime
The earliest known research or experiment in the United States that focused on
examining whether police patrols reduced violent crime was The Kansas City Preventive
Patrol Experiment (Wain & Ariel, 2014). The Kansas City Police Department conducted
an experiment from October 1, 1972 through September 30, 1973. The objective of the
experiment was to measure the effect that routine patrol had on the incidence of crime
and the public’s fear of crime (Wain & Ariel, 2014). The quasi experiment involved
three controlled levels of routine preventive patrol. The area termed reactive had no
preventive patrol. Police only came into this area when called by citizens for help. In
essence, this area had considerably reduced police visibility (Wain & Ariel, 2014).
The second area of the quasi experiment (Wain & Ariel, 2014), the proactive, had
police visibility increased to two or three times its usual visibility level. The third area
was the control, wherein the police maintained the same level of visibility. The three
areas, reactive, proactive, and control, experienced no statistically significant difference
in the levels of crime, citizens’ attitudes toward police services, citizens’ fear of crime,
police response time, or citizens’ satisfaction with the time police responded to calls
(Wain & Ariel, 2014). According to the Wain and Ariel (2014), the experiment showed
31
that routine preventive patrol in marked police cars had little value in the prevention of
crime or making the citizens feeling safe.
Wain and Ariel (2014) also reported findings regarding police attitudes toward
routine preventive patrol. Larson (1975) criticized the experiment because the
experimenters had not defined the concept of the traditional routine preventive patrol and
the vague concept the experimenters had was not explained to the officers in a
standardized way. Individual police officers had to decide the value and effectiveness of
the routine patrol through their own personally developed criteria. The police officers
interviewed were not certain as to the degree of routine patrol effectiveness. The
majority of the participants indicated that they would make changes if given the
opportunity, but some were accepting of the traditional routine as police officers. The
Kansas City Preventive Patrol Experiment (Wain & Ariel, 2014) was relevant to this
study because it is the earliest known experiment to test the assumption whether police
patrols could be effective in crime prevention and reduction. In spite of its mixed
reviews, the experiment laid the groundwork for further research on the subject.
Larson (1975) examined the design of the experiment to ascertain the conditions
under which the Kansas City Preventative Patrol conducted the experiment. Larson
questioned if the experiment maintained the conditions in the design in the execution of
the experiment. Larson further raised questions about the types of data the police
department researchers used in evaluating the findings. Because of the flaws in the
experiment’s design, execution, and documentation of findings of the Kansas City

32
Experiment, Larson concluded that the findings should not be used to extrapolate the
general value of a visible patrol presence.
Larson (1975) identified the following shortcomings in the Kansas City
Preventive Patrol Experiment:
1. The researchers did not make a description of the dispatch procedures in the
Kansas City Experiment. This description was important because it would
have played a key role in making a determination regarding how units spent
their committed time; besides, dispatch procedures affect many of the
performance measures that had a bearing on the experiment. A disregard to
describe the dispatch procedures affected the outcome of the findings.
2. There was no explanation for the manning levels for the experiment. This
lapse also made it difficult to compare the existing manning levels in the
experimental area to the manning levels during the experiment.
3. The researchers often relied heavily on performance measures, as perceived
by the citizens or participant observers. Though this approach was useful, the
actual data could have been more useful in some cases. A dearth of reliable
data described call-for-service volume, travel times, on scene service times,
and time-of-day variations.
4. Another flaw by the researchers was the participant-observers’ method of
choosing beats and cars. They favored those with a heavier workload that
33
biased the results in respect to travel time analysis, which could have also
biased the police-citizen encounters and the police noncommitted time.
Risman (1980) differed with Larson’s (1975) evaluation of the Kansas City
Preventive Patrol. Risman asserted that if visibility was a causal factor for any of the
dependent measures that Larson (1975) contested, the pairwise tests of significance
between two conditions—the levels of police presence in proactive and reactive beats—
should have yielded statistically significant results. The pairwise tests of significance are
the most liberal of procedures for analyzing the data collected.
Risman (1980) further disagreed with Larson that the negative results on nearly
all measured dependent variables, reported crime rates, official crime rates, victimization
surveys, and citizens’ perception of police presence, and service, were the effects of an
inadequate design. Risman further asserted that the negative results are indicative of
realistic variability in preventive patrol and that although Larson used his model to make
an approximation of mileage traveled per shift in the reactive and controlled beats, he
neglected to compute a parallel approximation of mileage for the proactive beats. An
explication of these neglected approximations of mileage for the proactive beats would
make it clear that patrol visibility in the proactive beats quadrupled the visibility in the
reactive beats, and doubled the visibility in the control beats. According to Risman, a
complete computation would show that the variability in preventive patrol within the
range of no-patrol to two or three cars per beat did not significantly affect crime rates or
citizen perceptions in the experiment.

34
Another early experiment to address the hypothesis as to whether police patrols
could have any significant effect on the reduction of crime was the Newark Foot Patrol
Experiment (1981). The impetus for the experiment was the desire by police
departments, policy makers, and citizens to know how to do more with fewer resources in
a situation where criminal activity has proliferated. Police departments in particular must
be efficient (Police Foundation Research Brief, 1981).
Foot patrol is one of the services frequently demanded by citizens, who often link
foot patrol with their feeling of safety in their neighborhoods (Police Foundation
Research Brief, 1981). Many citizens have expressed a feeling of safety with their close
contact with the police. Citizens also felt more secure when police were visible and were
patrolling the streets. The assumption as to whether or not police patrols did, in fact,
reduce crime needed testing.
Some police departments came to reject the function and viability of foot patrol as
antiquated, expensive, and even an anachronism (Police Foundation Research Brief,
1981). Other departments thought it was irrelevant to contemporary policing strategies.
Foot patrol also suffered because of the perception that foot patrol officers were low
status. Foot patrols were also perceived as a form of punishment for officers who
performed poorly. In 1973, however, the state legislation of New Jersey passed the Safe
and Clean Neighborhoods Act. One of the specifically mandated parts of this legislature
was the expansion of the presence and visibility of police protection, the foot patrol.
35
The design of the experiment demographically matched 8 feet patrol beats in
Newark. The foot patrol continued in four randomly selected beats, while it discontinued
in the 4 other beats. Foot patrols started in 4 beats where no foot patrols existed. The
governor of New Jersey invited the Police Foundation to make evaluations on the foot
patrol in 28 New Jersey cities with Newark selected as the primary evaluation site from
February 1978 until January 1979. The results indicated that introducing foot patrol
combined with other strategies was effective in building a perception of safety among the
people.
The Newark Foot Patrol Experiment (1981) was a development in the quest for a
proven assumption regarding the effectiveness of police patrols as a strategy for crime
prevention or reduction. The findings of the experiment further moved the debate
forward and subsequently researchers conducted further studies to get a clearer
understanding on the issue. The Newark Foot Patrol Experiment was also one of the
foundational studies during which researchers tried to learn about one kind of police
patrol and its effectiveness on crime prevention and reduction. However, the researchers
of the study did not find that police foot patrols reduced crime or violent crime (Newark
Foot Patrol Experiment, 1981).
Research Establishing the Effectiveness of Police Patrols
In spite of the position of some researchers that police patrols do not reduce
violent crime, a corpus of literature exists, asserting that police patrols do reduce violent
crime and crime in general. The Flint Foot Patrol Program (1981) in Michigan was one
36
of the early experiments that increased the knowledge of researchers and policy makers
regarding the efficacy of police foot patrols in decreasing crime or violent crime. As a
crime reduction strategy, the foot patrol in Flint also went beyond the stated strategy in
using foot patrol officers as part of a strategy in community problem solving. These
trained officers in the program were to do more than act as a viable deterrent to crime
(Andresen & Malleson, 2014).
Limited value of traditional patrol methods: Routine and random. The
concept of the Flint Foot Patrol was a result of the recognition by law enforcement that
the traditional methods of crime prevention, such as routine random patrol, had limited
value (Andresen & Malleson, 2014). The search for more effective methods of deterring
crime was evolving and the need for abandoning the traditional reactive methods of
fighting crime was urgent. Instead of requiring the police to work harder, they were
required to think harder and employ strategies that would make them more effective in
fighting crime.
The Flint Foot Patrol was part of a community-oriented policing strategy that
required a change in the balance of power between the police and the community
(Andresen & Malleson, 2014). The police alone under this model did not determine what
level of service to provide and the manner in which to provide the service. Both the
police and representatives of the community as a stipulation of the Flint Program made
decisions about police services.

37
Unlike the findings of the Newark Foot Patrol Experiment, which confirmed that
foot patrol had no effect on crime, the Flint Foot Patrol Program found that the foot patrol
brought about a 9% reduction in crime (Andresen & Malleson, 2014). The Flint Program
also found that the satisfaction that residents had for police services was significant. Of
the residents in Flint, 33% also knew the police officers by name, and 50% could
recognize the beat officer. The findings of the experiment brought about a resurgence in
the viability of police foot patrol as a useful police function.
Effectiveness of directed police patrols. The Indiana Project was one of the
experiments that established increased police patrols would reduce violent crime if they
were directed and not random (Andressen &Malleson). The other inference drawn from
the project was that specific crimes had to be the focus if police patrols were to reduce
crime. The project researchers suggested that, in accordance with the Kansas Gun
Experiment, the efficacy of increased police patrols was contingent on how the patrols
were conducted; the patrols would not be effective in reducing violent crime if they were
random but would be effective in reducing violent crime if they were directed at specific
crimes and areas (Andressen &Malleson)
In spite of the wide recognition of the foot patrol as a legitimate strategy in crime
prevention, the automobile continues to be the dominant mode and look of the police
patrol officer. Some researchers have also demonstrated that not all kinds of police patrol
can have an effect on violent crime or crime in general. For instance, the Kansas City
38
Prevention Patrol Experiment was one of the examples that demonstrated that non
directed police patrol would have no effect on reducing crime (Wain & Ariel, 2014).
The Kansas City Gun Experiment (1993) demonstrated that directed police patrols
could reduce violent crime, especially gun crime. The experiment developed out of a
federal grant, conducted in Kansas City, MO, at a time when handgun violence rapidly
increased in the city and the nation as well (Jiao, 2014; Wellford & Lum, 2014). This
national trend of gun violence that gripped the inner cities mostly resulted in high youth
homicide rates (Karandinos, Hart, Castrillo, & Bourgois, 2014).
Arguments among scholars who posited various reasons for the situation did little
to salvage the issue. Some scholars contended the deterrence power of having law-
abiding citizens carry guns. Other researchers asserted that more availability of guns
would result in increased gun homicides (Hemenway, 2014). The hypothesis implies that
the strengthening of existing laws about concealed guns is sufficient in reducing gun-
related crimes. The hypothesis needed to be tested to determine its validity.
The theory that ensued out of this program was that additional police patrols
would increase seizure of guns that could in turn reduce gun crime (Hemenway, 2014).
Deterrence and incapacitation were suggested by criminologists as possible mechanisms
for crime reduction. The assumption of the deterrence theory was that if the police seized
guns, illegal gun carriers would be less likely to carry them in the area. The
incapacitation theory, on the other hand, was of the assumption that if enough potential
39
gun criminals in an area had their guns seized, their ability to commit gun crimes would
diminish until perhaps the criminals were able to acquire new guns.
The project team knew from the start that confirmation of the hypothesis would
hardly demonstrate that gun seizures resulted in reduced gun crime (Hemenway, 2014).
They were cognizant of the fact that the design would not eliminate competing
explanations that could be suggested from the results. The team therefore settled for
finding an inverse correlation between gun seizures and reduction in violent crime, which
could suggest the value of further research and development (Hemenway, 2014). An
inverse correlation could also support a policy of extending the patrols.
The Pendleton Crime Reduction Project (2004) occurred at the beginning of
September 2004 and ended in June 2005. The project was a partnership between
Cincinnati City Council member, David Pepper, the Cincinnati Police Department, and
the Ohio Service for Crime Opportunity Reduction (OSCOR). The aim of the project
was to examine Cincinnati for crime patterns and the appropriate interventions given
what the data showed. The researchers conducted analyses for an area located in the
Pendleton neighborhood that was a relatively small neighborhood zoned for both
residential and commercial land use. Madensen et al. (2005) found that 2,104 calls were
made, and 37% originated within 500 blocks. The calls involved violence, disorder, drug
activity, displays of weapons, wanted suspects, and parking violations in Pendleton.
The key interventions that the OSCOR used and evaluated were traffic barricades,
directed police patrols, and Pendleton's Community Problem Oriented Policing (CPOP).
40
Madensen et al. (2005) stated that the directed police patrols that officers carried on bikes
and on foot brought about a decrease in crime during the patrols but that criminal activity
resumed after the directed patrol ended. The directed patrols could not continue because
of resource limitations. The CPOP teams that set up the barricades disbanded because of
rising tension with the Pendleton Community Council. The researchers were not able to
gather details to calculate the efficacy of this intervention strategy and little suggested
that the intervention had a significant effect on crime (Madensen et al., 2005).
Madensen et al. (2005) developed a comprehensive list of interventions to address
the open-air drug markets in Pendleton. High visibility patrol surveillance was ranked
highest among law enforcement activities to destroy the drug markets that were a source
of much of the violence. Madensen et al. suggested that high visibility patrol surveillance
that could be done by car, foot, bike, or horse, would be the most effective among the
other intervention strategies, but that the effect would be short-term because of resource
limitations and the inability of Pendleton to sustain the response for long periods of time.
Madensen et al. (2005) posited that even though directed police patrols could be
effective in crime reduction, such patrols would need to occur in conjunction with other
strategies. The researchers contended that police patrols cannot disrupt the drug markets,
which were a large source of violence alone but would need the partnership of
community members, leaders, city or county agencies, and other local organizations
(Madensen et al., 2005). The study was relevant to my study because of the issue of
whether increased police patrols could have a significant effect on crime or violent crime.
41
Madensen et al. believed that police patrols need to be supported by other intervention
strategies for success. This project also explicated the problem of resource limitations
that could affect the sustained implementation of police patrols to fight crime (Madensen
et al., 2005). Police patrols on the streets could be the best intervention strategy to fight
crime or reduce crime but a city, town, or jurisdiction may not have the resources to
sustain the intervention.
The objectives of the researchers of the Kansas City Gun Experiment (1993) were
to decrease crimes resulting from the use of guns. When the Police Foundation evaluated
the experiment, it found that directed police patrols in gun crime hot spots reduced gun
crimes by increasing the seizures of illegally carried guns. The following were specific
findings of the evaluation of the experiment: (a) gun seizures increased by 65% resulting
in a 49% decrease in gun-related crimes, (b) gun crimes and seizure did not change
without patrolling, (c) homicides were reduced only within the target area, (d) citizens
were less fearful about crime-related incidents within the target area, (e) traffic stops
were the most useful method of gun seizures, and (f) two-thirds of arrests were made
outside the target area (Madensen et al.). Unlike the Kansas City Preventive Patrol
Experiment (1981) that was a random patrol, the Kansas City Gun Experiment was a
directed patrol and proved to have a significant effect on the reduction of gun violence.
Rosenfeld, Deckard, and Blackburn (2014) and Hart and Miethe (2015) found that 100%
more directed police patrols produced less hot spot crime.

42
Role of foot patrol in crime reduction. The Philadelphia Foot Patrol
Experiment (2009) challenged the long held opinion by many police departments and
criminologists that police foot patrols did improve community perception of police and
reduce fear of crime but did not reduce crime. A research collaboration between the
Philadelphia Police Department and researchers in the Department of Criminal Justice at
Temple University in some of Philadelphia's most violent areas produced findings that
could bring about revisions of the traditional position that police patrols do not reduce
crime (The Philadelphia Foot Patrol Experiment Research Brief, 2010).
The researchers of the Philadelphia Foot Patrol Experiment Research Brief (2010)
selected 120 foot-beats from an identified potential of 129 for the experiment. The
researchers ranked the foot beats by the weighted volume of violent crime and paired
with a foot beat of a similar rate in crime (The Philadelphia Foot Patrol Experiment
Research Brief, 2010). The selection of beat from each pair was determined in a random
way. One became the target beat with the other as the control beat. The officers
generally executed their patrols in pairs. The patrols commenced from Tuesday morning
to Saturday night in two shifts, 10:00 am to 6:00 pm and 6:00 pm to 2:00 am. The
researchers gave the police officers in the foot patrol area an initial intelligence brief.
Within the targeted area, the results indicated: (a) a decrease in crimes of 20%, (b) a 12%
decrease in crimes involving vehicles, (c) a 28% decrease in crimes involving drugs, (d) a
51% increase in traffic enforcement, and (e) a 13% increase in arrests (Hart and Miethe).
43
The researchers of the Philadelphia Foot Patrol Experiment Research Brief (2010)
discussed the critical issue of crime displacement as a possible factor for the reduction of
crime in the patrolled beat. Because some target areas were close to others, they had to
combine the areas to examine the issue of crime displacement. The researchers
concluded that some displacement occurred but not as much when compared to the target
areas (The Philadelphia Foot Patrol Experiment Research Brief, 2010). Police patrols
prevented 85 crimes in the target areas, offset by a 35 crimes increase taking place in the
displacement areas that immediately surrounded the target areas. The results indicated
that displacement occurred but the net benefit gained in the patrolled area in reducing
crime was more than the percentage of displacement that occurred (The Philadelphia Foot
Patrol Experiment Research Brief, 2010).
The findings of the Philadelphia Foot Patrol Experiment (2010) were contrary to a
similar experiment in Newark in 1981 that did not find foot patrol to have an effect on
crime reduction or prevention. The Philadelphia Foot Patrol was relevant to this research
study because of the demonstration of the improved research design and methodology,
and the hypothesis that increased police patrols can deter or reduce crime. Unlike the
Kansas City Preventive Patrol Experiment and the Newark Foot Patrol Experiment that
were random, the Indianapolis Directed Patrol Project (1997) targeted two areas, the
north target beats that were approximately 3 square miles with about 16,000 residents.
These beats were predominantly African American and low income. The east target
beats covered an area of 1.7 square miles with approximately 14,000 residents. The east
44
beats had mostly White residents and a small percentage of Hispanics, and the area was
also low income (Corsaro et al., 2012; Hoelzer & Gorman, 2011). The north target area
had a homicide rate three times that of the city, and robbery and aggravated assault rates
were almost twice that of the city (Corsaro et al., 2012). Property crime rate was slightly
lower than that of the city. The homicide rate of the east area was between that of the
north area and that of the city. The rate of robbery and aggravated assault was about
twice that of the city. The east side unlike the north side had a higher rate of property
crimes.
The east and north had slightly different strategies to their directed patrol
initiative (Corsaro et al., 2012). The east increased the number of traffic stops in a more
targeted fashion to increase the investigations of those people the patrol considered
suspicious and focused on the seizure of illegal weapons and drugs (Corsaro et al., 2012).
The east applied a general deterrence approach. Police increased motor vehicle stops to
maximize a sense of increased police presence. The police in the east side stopped
vehicles for any type of motor vehicle offense, such as an expired license plate or a
taillight out. This approach was based on the assumption that increased police presence
would deter offenders (Corsaro et al., 2012). The patrol also anticipated that the large
number of vehicle stops would yield seizures of illegal weapons and drugs. Unlike the
east, the north district followed a specific deterrence or targeted offender approach that
involved more vehicles for inspection (Corsaro et al., 2012). The police patrol was more
likely to stop cars that drove slowly through known drug-selling areas and those cars that
45
exhibited a traffic infraction, or sped through the neighborhood. The north patrol
expended resources on suspected individuals and illegal behavior to improve seizure of
guns (Corsaro et al., 2012).
Police Numbers do Matter in Crime Reduction
Gill et al. (2014) and Vollaard and Hamed (2012) asserted that increases in the
number of police have a positive correlation, which differed from the assertions of Kleck
and Barnes (2014) and Weisburd, Hinkle, Braga, and Wooditch (2015), who stated that
police increases on the streets have insignificant effects on crime reduction. Among the
four factors Levitt stated could explain why crime fell in the 1990s was increases in the
number of police. Weisburd et al. doubted the correlation between a decline in crime and
an increase in the number of police. The political response to increases in crime was to
hire more police, which translated that the number of police affected crime rates. The
endogeneity problem affected the clarity of the theory, until recent researchers addressed
the problem. Marvell and Moody (1996) used the "Granger-Causality" approach on
panel data to demonstrate that increases in the number of police were associated with
reduction in crime in the future.
McElvain, Kposowa, and Gray (2012) found that increases in police or police
patrols did not decrease crime. In another attempt to address the endogeneity problem,
Corman and Mocan (2000) used high frequency time series data to reduce the problem.
These researchers also used the approach for New York and found that elasticity ranged
from -0.29 to -1.385 across crime categories with a median value of -0.452.
46
Levitt (1997) used the timing of mayoral and gubernatorial elections as a tool for
the hiring of police. As politicians disproportionately increased the size of the police
force before elections, and because elections were unlikely to affect crime directly,
elections became a plausible instrumental variable. Through the strategy, elasticity
estimates ranged between -0.05 and -1.98 across the crime categories with a median
value of -0.79. McGary (2002), however, pointed out that the correctly computed
standard errors made the point estimates statistically insignificant. Biderman (2012)
stated a more precise estimated elasticity of -0.43 to -0.50.
The FBI tracked the number of police officers per captain in the UCR and
reported that the increase of 50,000–60,000 police officers, approximately 14%, could
account for crime reduction of 5–6% across the board (Biderman, 2012). Levitt's (1997)
contribution to the debate about the negative effect of the increased number of police on
the streets seemed to put to rest many doubts by addressing the endogeneity problem that
led some researchers or criminologists to doubt the positive effect of increased police
presence on the beats of a city.
In another evaluation study by Zimmring (2012), I asserted that police did matter
in crime prevention. In the evaluation article, the researcher stated that the arguments
that crime drops were because of socioeconomic factors were inaccurate (Zimmring,
2012). The researcher stated that police interventions were instrumental in the decrease
of crime rates. The researcher posited that police could deter or prevent crime by their
presence and by persuasion, thus reducing the opportunities for crime and arresting
47
wrongdoers. Zimmring disputed the theorists who posited that crime was the result of a
collection of root causes, such as social injustice, racism, and poverty. The implication
of the root cause theory did not place preference on the use of police to fight crime, but
rather emphasized that crime would decrease if society addressed the root causes of
crime. The root causes theory entails that police have little influence on crime, because
they react to crime after it occurs. The root causes theory de-emphasized the role of
police in crime prevention (Zimmring, 2012).
To validate the effect of police on society, Jenkins (2015) analyzed how the police
were able to decrease crime in New York City. The researcher isolated three innovative
approaches. The first approach was the idea of problem solving as a key role prevention,
advanced by Professor Goldstein (1979). The second approach was the broken windows
idea formulated by Professor Wilson (1982). The third approach was the emphasis on a
new way of managing police resources and tactics, known as comp stat. The police
department fused these three innovative ideas to enable the police to prevent crime in
New York City.
A major contribution of this evaluation to the literature review is the position
taken by the police department of New York, believing that police could have a negative
effect on crime reduction or prevention if they focused on problem solving and ensured
that they fixed any broken thing that could give rise to further crime (Jenkins, 2015).
Police would also be required to follow a new method of managing police resources and
tactics to succeed in crime prevention or deterrence. The evaluation qualified how much
48
police could matter if law enforcement implemented the ideas of problem solving, broken
windows, and the new ways of managing police resources and tactics. If the police
implemented these ideas, they could matter in crime prevention.
In this evaluation article, Jenkins (2015) did not address police patrols by
themselves but rather addressed how the ideas in comp stat, broken windows, and the
innovative management of police resources and tactics could help police patrols to be
effective. The researcher asserted that during the last decade, crime rates in New York
plunged. Murder declined by more than 70%, robbery by more than 60%, total violent
crime offenses by more than 50%, and total property felonies by more than 60%
(Jenkins).
Criminologists understood a key difference between this study and earlier studies,
involving police patrol effectiveness research in terms of the importance of police in
crime prevention or deterrence in the following four approaches (Jenkins, 2015). First,
most researchers examined crime data but few examined crime, in relation to other
factors. Second, while most researchers used homicides as the only outcome measure to
examine factors that influenced crime rates, this research included an outcome variable
that captured a wider spectrum of criminal activities: murder, rape, felonious assault, and
robbery. Third, this research avoided the disadvantages of unsuitability comparison
when using New York and the tendency not to compare New York at all. The
researchers simulated the comparison groups. They used New York as 76 separate cities
corresponding to 76 police precincts. They then viewed the variations economically,

49
demographically, and socially with police interventions in unison from 1993 to 2000.
Fourth, this research differed from other analyses on the subject because Jenkins (2015)
did not draw data solely from secondary sources to generalize about police practices.
Some researchers have drawn from secondary sources without familiarity with the
dynamics of police departments, the style and nature of the departments' police tactics,
the methods of data collection, and the overall police departments' practices in the
context of where the practices are rooted. The researchers ensured that they collected
original data as well. The researchers posited that police did matter in crime reduction, as
evidenced in New York in the 1990s, but that if the police were to affect the reduction of
crime of any kind, patrols or other strategies had to be focused on a problem and
concentrate their resources on hot spots where crime is concentrated.
Hot-Spot Policing
Braga (2006) conducted a meta-analysis of randomized experiments on the
effectiveness of police patrols and other strategies to reduce crime, and posited that the
experiments revealed statistically significant mean effects sizes favoring hot spot policing
interventions. Braga measured for immediate spatial displacement and found limited
results. One of the focused interventions that contributed in reduction of crime was the
use of directed patrols in policing the hot spot areas. The other strategies used were
proactive arrests and problems-oriented policing. The hot spots perspective in Braga's
meta-analysis posited that police could reduce crime by focusing their limited resources
50
on the small number of places that often generate the majority of crimes, as asserted in
Rosenfeld et al. (2014).
The development of technology has helped in crime prevention (Byrne & Marx,
2011). The police's use of crime mapping technology and computerized database studies
revealed to police departments that crime clusters were usually in discrete small places,
street blocks, apartments, and other specific addresses responsible for a high percentage
of the total crime. The Minneapolis Hot Spots Experiments showed that about 5% of the
addresses generated about 50% of all the calls from citizens for service (Rosenthal et al.).
The issues of displacement and diffusion effects were addressed. Five studies
examined spatial displacements of crime into immediate surroundings of the targeted hot
spots. None of the five studies reported significant or substantial immediate spatial
displacement. In four of the studies examined, Braga (2006) found possible diffusion
effects associated with the focused police interventions. Johnson, Guerette, and Bowers
(2012) posited that researchers should also focus more on crime diffusion than crime
displacement, even though hot spots with focused police strategies, such as directed
patrols, may experience reported crime reduction.
In using terror alert levels to estimate the effect of police on crime, Klick and
Tabarrock (2005) posited a credible estimate of the causal effect of police on crime, in
spite of the need for more research for scholars and criminologists to determine the effect
and magnitude for a generalization to other cities besides Washington, DC. Klick and
Tabarrok confirmed the findings of Biderman (2012) that increased police on the streets
51
could have a significant influence on crime reduction. McCray (2002) disagreed with
Biderman and argued that a programming error made Biderman’s results to appear more
precise that justified. McCray posited that only a stronger research design or perhaps
heroic data collection could make researchers avoid making imprecise estimates of the
causal effect of police on crime.
Klick and Tabarrock (2005) claimed to have developed a stronger research design
than those utilized in similar research. The researchers posited that a new data source
enabled them to make better estimates of the causal effect of police on crime (Klick &
Tabarrock, 2005). Klick and Tabarrock used the high-alert times in DC and the police
increase on the streets to break the circle of endogeneity and estimate the effect of police
on crime. Many researchers have used the endogeneity problem to criticize the previous
findings on the subject (Biderman, 2012).
Klick and Tabarrock (2005) addressed another critical issue better than previous
research by using improved data. Many previous studies involved annual data subjected
to an inherent trade-off, for instance a longer time series would improve the precision of
estimates but would increase the possibility of omitted variables bias. Panel data, on the
other hand, reduces the need for a long time series but raises the problem of endogeneity
and omits variable bias in the cross sectional component. Klick and Tabarrock (2005)
focused on a single city, Washington, DC, reduced the omitted variable bias in the cross
sectional component.
52
Klick and Tabarrok (2005) also used the easily identifiable and clearly exogenous
shock to examine the effects of police on crime. The importance of this research for the
study was the treatment the researchers gave to the terror alert level that was turned on
and off repeatedly during their sample that enabled the researchers to solve the problem
of endogeneity. The researchers also used daily crime data that made the treatment
window short (Klick & Tabarrock, 2005). The treatment window lasted days and weeks,
making the results less likely to be because of other factors and enhancing the credibility
of the findings. Klick and Tabarrock showed that a 50% increase in police presence led
to 15% reduction in crimes.
Klick and Tabarrok’s (2005) findings showed a marked difference from Levitt
(1997), because they were able to show a credible estimate of the causal effect of police
on crime. The difficulty posed by Klick and Tabarrock, however, was to determine if this
effect and its findings could be applicable for other cities. The researchers posited that in
principle, their design that used terror alert changes as exogenous shocks to police
presence was good and that other researchers could use daily crime data in analyses of
crime patterns in other metropolitan areas to make a determination of the effect of
increased police or police patrols on crime (Klick & Tabarrock, 2005). The findings in
the research showed a marked difference from the research previously discussed. Klick
and Tabarrok stated that violent crimes showed no response to increased police patrols on
high alert days. The large declines in crime were in automobile thefts and thefts from
automobiles.
53
Another major difference between previous research and the research of Klick
and Tabarrok (2005) on the influence of more police patrols on crime was the focus on
the percentage by which increased police patrols can diminish crime. The researchers
decided to focus on cause and effect and assumed that if increased police patrols could
reduce crime, they wanted to know the size of the effect of the reduction or deterrence
(Klick & Tabarrok, 2005). Klick and Tabarrok determined the percentage by which
police patrols could deter crime in terms of costs and benefits to ascertain whether
employing more police for patrols would be beneficial to a city. Police jurisdictions
spent a lot of money in previous years and it would be worthwhile for them to determine
whether taxpayers' money was spent well by calculating the benefits against the costs.
Klick and Tabarrok (2005) also posited that their research was a natural
experiment. The researchers contended that to separate cause and effect and to determine
whether more police on the streets would reduce crime would require a natural
experiment, an event that changed the number of police officers for reasons that had
nothing to do with the crime rates (Klick & Tabarrok, 2005). Klick and Tabarrok
compared the crime rates before and after the terror-alert to determine the effect. Using
the changes in the terror-alert level provided the kind of natural experiment they needed,
because the shifts in police deployment were large enough to made effects easier to
determine. The number of police officers on patrol fluctuated during days and weeks,
rather than months or years. The authors did not observe any displacement because
crime decreased throughout the city, Washington, DC (Klick & Tabarrok, 2005). Klick
54
and Tabarrok posited that other researchers could replicate their research in other cities
since the terror alert system operated nationally.
Groff et al. (2013) conducted a study to determine causal effects of more police
on crime. Marvell and Moody (1996), Ratcliffe and Breen (2011), and Kleck and Barnes
(2014) found either a positive effect of police presence on crime or no relationship
between the variables, police and crime. These findings were contrary to Becker (1968),
whose theory contended that crime would decrease when police presence increased.
Addressing the Endogeneity Problem
The researchers who found a positive effect of police presence on crime faced the
endogeneity problem because of the identification of crime and the presence of the police
(Kelman, Hong, & Turbitt, 2013). A likelihood existed that a city would hire more police
officers because the crime rate increased. Areas beset by high crime would have more
police than areas with low crime rates; such a situation would introduce a positive bias in
the police coefficient in a crime regression. Researchers in the field of study had the
central challenge of breaking the endogeneity problem if they were to isolate causal
effects of police on crime (Groff et al., 2013). The researchers presented a different
approach from Levitt (1997) to estimate the causal effect of police on crime.
When terrorists attacked a synagogue in Buenos Aires, where 85 people were
killed and 300 were wounded, the federal government assigned police protection to every
Jewish and Muslim building in the country. Groff et al. (2013) presumed the institutions
exogenous in a crime regression because of the geographical distributions and the terrible
55
crime to constitute a natural experiment in which the simultaneous identification of the
crime and the presence of the police could be broken. The researchers found that city
blocks that received police protection experienced significantly fewer car thefts than the
rest of the neighborhoods (Groff et al., 2013). The effect was large. Car thefts fell by
75% in the blocks with the protected institution. The researchers, however, concluded
that the effect was local as they found no evidence that police presence in a given block
reduced car theft one or two blocks away from the protected buildings (Groff et al.,
2013). Groff et al. also posited that their estimates were not appropriately interpreted as
the causal deterrent effect of police patrols on car crime. The experiment of Groff et al.
(2013), unlike others previously discussed, did not address the problem of the
displacement of crime. The researchers argued that it was possible that car thefts were
displaced in a way they were not able to measure (Groff et al., 2013). The effect of
police on crime could have been smaller than their estimate suggested.
The experiment of Groff et al. (2013) is critical to the research because the
researchers broke the endogeneity challenge, which was a barrier in finding a negative
causal effect of police on crime. The findings were similar to the research of Klick and
Tabarrok (2005), because both occurred in a naturalistic setting. Levitt (1997) developed
a different approach to break simultaneity by using instrumental variables, which McCray
(2002) criticized. Groff et al. decided to use a natural and randomized experiment. This
experiment strengthened the position of those scholars who posited that more police on
the streets had a negative effect on crime. Whereas Levitt found a large negative and
56
significant effect of police on violent crime, Groff et al. found a negative effect of police
on car thefts or property crime.
Among the researchers who found a relationship between police levels and crime
were Kovandzic and Sloan (2002). Kovandzic and Sloan found that prior researchers
who posited that police levels had little effect on crime rates were methodologically
flawed. The researchers examined whether police levels had any effect on crime rates.
Kovandzic and Sloan subjected the data to a multiple time series design and revisited the
police crime relationship using a sample of large cities. The study results revealed that
increased police levels reduced most types of crime at the county level (Kovandzic &
Sloan, 2002). The findings of Kovandzic and Sloan’s research challenged the assertion
of Cameron (1988), who reviewed 22 publications on the relationship between police
levels and crime.
Marvell and Moody (1996) also reviewed 36 published studies in which prior
researchers examined the effect of police levels and crime rates. The analysis confirmed
that in 10 of the 36 studies, the researchers found an inverse relationship between police
levels and crime rates; as police levels increased, crime rates decreased (Marvell &
Moody, 1996). The relatively consistent findings by prior researchers strengthened the
theory that no relationship existed between police levels and crime rates. Critics argued
that police were not directly involved in the handling of crime (Bayley, 1985, 1994;
Bensen, 1984).
57
O’Donnel (2011) posited that the assumption that more police decrease crime
justifies the hiring of additional police and the building of more prisons. In their
research, Kovandzic and Sloan (2002) disagreed with the assumption posited by some
researchers that increased police levels did not decrease crime. The researchers also
disagreed that police departments or policy makers believed that increased police or
police patrols decreased crime to justify the hiring of more police (Kovandzic & Sloan,
2002). Kovandzic and Sloan criticized prior researchers for their inability to mitigate
specification problems, including simultaneity and the incomplete control variable.
Marvell and Moody (1996) presented a more appropriate methodology that solved the
specification problems that led prior researchers to assert that no relationship existed
between increased police levels and crime. Kovandzic and Sloan (2002) used the
Granger test to explore causal direction and to determine the ability of one variable to
predict another. Researchers usually use the Granger test when they were unable to
specify the structural model.
Specifically, Kovanadzic and Sloan (2002) used the Granger test to examine
causal effect, as the test identified instantaneous causation because the value of the
current year of the independent variable serially correlated with its 1-year lagged value.
The researchers accomplished this by regressing police levels on lags of themselves and
on lagged crime rates (Kovanadzic & Sloan, 2002). The results of the research showed
strong evidence that increased levels of police patrols led to lower crime rates. The
results showed that a 10% increase in police levels lowered crime rates by 1.4% over
58
time (Kovanadzic & Sloan, 2002). The findings of Kovandzic and Sloan were consistent
with the findings of Marvell and Moody (1996), who suggested that local government
leaders make decision regarding the firing of police officers based on the overall
fluctuation in crime rate than in specific fluctuation data. The findings of the research
advanced more evidence that statistical models in prior police crime research suffered
from simultaneity bias. The simultaneity bias that prior research did not address with
appropriate procedures led many researchers to believe that increased police levels had
no effect on lowering crime.
Telep and Weisburd (2012), in a review, asserted that increased police levels
decreased or deterred crime. Telep and Weisburd stated that deterrence effect was one of
the rationales behind the proposition that police could reduce crime. Becker (1968)
popularized the proposition of deterrence that Beccaria and Jeremy Bentham introduced.
From a basic perspective, police presence would deter crime because it raised the
probability of the offender being caught. Telep and Weisburd disagreed with prior
researchers who posited that little or no evidence demonstrated that more police could
reduce crime. Telep and Weisburd criticized those studies for failing to address the
endogeneity problems. Marvell and Moody (1996), Nagin (1998), and Levitt (2002a,
2002b) have stated that the positive correlations that some researchers found may
probably be the result of confusion between correlation and causation.
Researchers needed instrumental variables that correlated with police presence
but uncorrelated with the error term that must affect crime only through the presence of
59
police (Levitt, 1997). Telep and Weisburd (2012) estimated the effects to show that
changes in state taxes had an effect on the levels of local police as illustrated in Figure 1.
Figure 1. The effects on local police from changes in state tax rates.
Telep and Weisburd (2012) compared the time series data on crime rates, police
numbers, and sales tax rates which showed that significant decreases in both violent and
property crimes after 1990 correlated with the increase in police and sales taxes. Telep
and Weisburd’s research findings further reinforced the position of researchers who
maintained that prior researchers failed to find a correlation between increased police
levels and crime reduction because they did not address the problem of simultaneity
(Kovandzic & Sloan, 2002; Levitt, 2002 a; Marvell & Moody, 1996; Nagin, 1998).
Through a British Crime Survey, Vollaard and Hamed (2012) found that
increased police patrols had a significant effect not only on property crime, but on violent
crimes as well. In the survey, the researchers posited that the crime data from the survey
did not suffer from measurement errors that earlier research suffered from, especially
from the propensity by researchers to record violent crime that was susceptible to
changes in the number of police than of police statistics on property crime (Vollaard &
Hamed, 2012). In their analysis of the collected data, Vollaard and Hamed found that the
60
estimated effect of police on violent and property crime was similar in size as a 1%
increase in police resulted in a 0.7% decrease in crime. The findings are similar to those
in earlier studies in the United States. The decrease of crime in the 1990s in the United
States was partly because of the increase in the number of police or police patrols
(Donahue & Ludwig 2007; Gill et al., 2014). Previous researchers, including Corman
and Mocan (2000) and Klick and Tabarrok (2005), also found a significant effect of
increased police patrols on property crime.
While some researchers found a significant negative effect of increased police on
property crime, Vollaard and Hamed (2012) found a significant negative effect of
increased police on both property crime and violent crime. The researchers posited that
the insufficient evidence in previous studies was based on the data collection methods.
Vollaard and Hamed posited that property crimes tended to be recorded more often than
violent crimes, because citizens more often report property thefts to the police than
crimes of violence. Besides the violent crime of homicide, police less frequently
recorded a violent crime than property crime. Incomplete police records, victim
reporting, and police recording biases compounds the effect of police in crimes. Vollaard
and Hamed stated that the inability of prior researchers to address the reporting and
recording biases between property crimes and violent crimes led to the conclusion that
increases in police had no effect on violent crime (Vollaard & Hamed, 2012).
In their research, Vollaard and Hamed (2012) addressed the measurement error
that resulted in biased estimates. The approach made a difference that resulted in finding
61
a significant negative effect on violent crime by police increases, which was relevant to
this study. The researchers based their analysis on data at police force area levels.
Vollaard and Hamed posited that robust effect of increased police on violent crime did
not occur in the studies of Corman and Mocan (2000, 2005) and Levitt (2002). The
research was critical to my study because it emphasized the importance of addressing the
measurement error that was central and ignored by prior researchers when they studied
the effect of increased police or police patrols on crime. Vollaard and Hamed used crime
survey data rather than relying primarily on police recorded data for the same set of
police force areas and for the same time. The British Crime Survey became one of the
recent emerging studies that provided evidence that the effect of increased police or
police patrols had a significant negative effect on violent crime as well as on property
crime (Vollaard & Hamed, 2012).
Draca et al. (2008) used terrorism-related events to examine the police-crime
relationship, because terror attacks would often bring about increased police presence in
identified places. The researchers considered the police-crime relationship before and
after the terror attack in a metropolitan area (Draca et al., 2008). The study was similar to
that of Groff et al. (2013) and Klick and Tabarrok (2005), because they also used
terrorism-related events in their research. The application of the research was, however,
more general than the previous two studies by Groff et al. and Klick and Tabarrok. Draca
et al. covered a large metropolitan area, and the scale of the attack provided the
researchers a valuable setting to examine the relationship between police and crime; it
62
provided a natural setting that the other two researchers had. The researchers also had
accurate data on police deployment that they used to identify the magnitude of the causal
effect of police on crime (Draca et al., 2008). Draca et al. were able to use the crucial
aspect of the "natural" setting to identify a causal effect and to establish the exclusion
restriction.
The findings of Draca et al. (2008) were different from previous researchers
(Kleck & Barnes, 2014; Marvell & Moody, 1996; Ratcliffe & Breen, 2011), who did not
find any relationship between increased police or increased police patrols and crime
reduction. Draca et al. posited that the failure in prior research to find a relationship
between increased police or increased police patrols and reduction in crime was because
of those researchers’ inability to identify the direction of the relationship between the two
variables. Draca et al. posited that police increases had negative effects on property and
violent crimes, finding that the scale of police deployment during the terror attack was
greater than the highly localized responses previously studied. The police keeping their
force levels constant had a large effect on crime. Clear evidence of the timing and
location of reductions in crime coincided with the increase in police deployment (Draca
et al., 2008). The findings were contrary to Hart and Miethe (2015), who argued that no
matter how police were deployed, police presence would not deter crime. The increase of
police patrols by 30% in the 6 weeks following the July 7, 2005 bombing made crime fall
significantly in other areas of the city under investigation (Draca et al., 2008). Draca et
63
al. estimated an elasticity of crime with respect to police of about -.3, so that a 10%
increase in police activity reduced crime by about 3%.
The researchers did not find strong evidence of temporal or spatial displacement
effects because of the 6-week police intervention. The research was relevant to this study
because the researchers showed that an increased number of police or increased police
patrols on the streets of an urban area could reduce property crime and violent crime.
The researchers found that increased police patrols could reduce crime and addressed the
problem of endogeneity that posed difficulties in establishing a relationship between
police and crime, as found in the prior studies (Marvell & Moody, 1996; Ratcliffe &
Breen, 2011).
Vollaard (2005) observed that many policy makers based their decisions about
police budget primarily on anecdotal evidence regarding the effects of more police in
crime rather than the results of systematic research. Vollaard analyzed the effect of
additional police on the chance of becoming a victim of crime and nuisance during 1996–
2003 in a metropolitan area. After the analysis of changes in police personnel and trends
in crime and nuisance, the results indicated that: (a) police patrol is more effective in
urban areas, (b) increase in the number of police decreases crimes, and (c) more police in
an area might lead to criminals fleeing to other places (Vollaard, 2005).
Vollaard's (2005) based the approach to isolate the effect of rising police levels
from many other factors on the premise that more police personnel decrease the rate of
crime. Vollard focused on the growth rates of police and not the historically grown levels
64
of police and crime in a region. The researcher also focused on the differences in police
personnel growth between regions and estimated the crime trends in those regions.
Vollaard used the approach or design to identify the effect of police on crime. The
researcher avoided the use of the budget formula that could allow a third a third factor to
obscure the effect of police on crime (Vollaard, 2005). Policy makers and law
enforcement distribute most of the police budget among police forces using the budget
formula. Vollaard studied various characteristics, such as number of shops and length of
streets, to determine the relative need for police resources. If the budget formula shifted
the growth in police resources to those areas with the highest crime trends, a strong
possibility would exist to underestimate police effectiveness, because in such a situation,
police resource problems would lead to a positive relationship between growth in police
personnel and increases in crime.
By analyzing the events from 1996–2003, Vollaard (2005) conducted an
uncontrolled experiment to answer the basic question about police effectiveness. The
researcher focused on whether the decline in crime was the result of police or better labor
market opportunities. Vollaard found that the marked expansion in police personnel took
place concurrently with the aggregated crime trends. Police forces increased by 13,000, a
growth of about 30% (Vollaard, 2005).
Braga, Hureau, and Papachristos (2014) evaluated the Operation Ceasefire, Los
Angeles (2000) and found that the use of increased patrols in the area of Hollenbeck
where violent gangs were most active brought about a reduction in violent crime. The
65
use of increased police in Boston and Los Angeles was successful in spite of the
geographical and racial differences (Braga, Hureau, & Papachristos, 2014). In Boston,
the gangs that perpetrated the violence were mostly African American; in Los Angeles
those responsible for the violence were mostly Latino. The gangs in Los Angeles were
bigger and more entrenched than in Boston (Braga et al., 2014). Another notable
difference between these two cities lay in the population difference. Los Angeles had a
population of 3,849,378 (U.S. Census Bureau, 2006) and Boston had a population of
589,141 (U.S. Census Bureau, 2000). In spite of the differences in population, ages of
gangs, and entrenched levels of violence in both cities, increased police patrols succeeded
in reducing violent crime in Boston, such as homicide (63%; Braga et al., 2014).
In Los Angeles, increased police patrols and intensive law enforcement in the
selected five police reporting districts of Boyle Heights where the gangs, Cuatro Flats and
TMC, were most active, saw a decline in the suppression stage, but the decline was
stronger in the deterrence stage by 37% overall and 24% in the rest of Boyle (Tita et al.,
2005). The comparison area outside of the five districts that did not have increased
police patrols only had a 3% decrease in violent crime. Operation Ceasefire (2000)
provided relevance to this study because it affirmed that increased police patrols in a city
can reduce violent crime, contrary to some researchers argued that increased police
patrols do not reduce violent crime. Tita et al. (2005) demonstrated that the size of the
population and the entrenched nature of violent crime in a city do not deter success if the
operations of law enforcement are focused.

66
Other Factors That Affect the Rates of Violent Crime
Unemployment
Unemployment has long been one of the factors associated with crime
(O’Donnell, 2011). Becker (1968) and O’Donnell (2011) found that the effect of
unemployment on crime is quantitatively significant. Papps and Winklemann (1999)
explored the contentious issue and found some evidence of significant effects of
unemployment on crime. The effects were for both total crime and for some
subcategories of crime. Box (1987) reported that out of 35 reliable studies that dealt with
the topic, researches of 20 studies found a positive relationship between unemployment
and crime; the others found no relationship. Small and Lewis (1996) found strong
support that crime and unemployment are linked, with unemployment causing crime.
The type of data prior researchers used to obtain the empirical literature on the topic for
was based on aggregate time series data and aggregate cross-section data, which have
affirmed a causal relationship between unemployment and crime (Brenner, 1978;
Chapman, 1976; Levison, 1976; O’Donnell, 2011). Entorf and Spengler (1998),
however, found that unemployment had a small and often insignificant relationship by
using a regional panel study. When Papps and Winkelmann revisited the research
regarding the relationship between unemployment and crime, they used regional panel
data, regressing crime rates on unemployment rates by using fixed and random code
effects. The researchers found a causal relationship between unemployment and crime
(Papps & Winkelmann, 1999).

67
Education
Another factor considered to have an effect on crime is education. Machin,
Marie, and Vujic (2011) found that schooling significantly reduced the probability of
incarceration. The researchers concluded that the differences in educational attainment
between Black and White men is an explanation for the 23% gap of the Black-White
incarceration rates after analyzing the FBI data on arrests for different types of crime
(Machin, Marie, & Vujic, 2011). Machin et al. also found that the biggest effects of
education on crime were linked to murder, assault, and motor vehicle theft.
Machin et al. (2011) delineated a number of factors, such as patience, to acquire
an education, development of positive social networks, moral values taught in schools,
and the probability of higher wages from acquiring labor market skills through education
to contribute to a correlation between education and crime. This analysis centered on a
human capital approach; that an accumulation of human capital during a lifecycle would
explain why those adults who had developed formal labor market skills would be averse
to criminal activity, especially for violent crimes that usually carry a longer incarceration
period (Machin et al., 2011). Gould et al. (2002) also estimated the effects of low skill
wages on violent crime to be larger than for property crime. Lochner (2007) stated that
higher wages raise the opportunity costs because the increase is more costly for persons
with better education or better labor market opportunities to engage in criminal activity
than those with little or no education or labor market opportunities.

68
Some researchers do not agree to a causal link or any significant relationship
between education and crime. Kleemans (2012) asserted that neither years of schooling
nor the receipt of a high school diploma had any significant effect on somebody's
criminal activity. Tauchen et al. (1994) also found no significant relationship between
education and crime.
Homeownership
Homeownership is another factor believed to have a negative effect on crime. Ni
and Decker (2009) used county-level census data and Uniform Crime Reports for the
United States to determine the influence that homeownership had on crime rates. The
researchers controlled for income's effect on homeownership rates and found that
homeownership had a strong and statistically significant negative effect on property
crime and violent crime (Ni & Decker, 2009). Ni and Decker asserted that increases in
homeownership rates brought about a reduction of criminal activity. Ni and Decker also
stated that with estimates putting the cost of fighting criminal activity at more than 5% of
U.S. GDP (Cooter & Ulen, 2000), it would be desirable to pursue the promotion of
homeownership because of its negative effect on crime.
Cost of Living (Inflation)
Researchers have ascertained inflation to be one of the determinants of aggregate
property crime (Seals & Nunley, 2007). Rosenfeld (2014), Torruam (2014), and
Tamayo, Chavez, and Nabe (2013) found a causal relationship between inflation and
crime that could be found in the crime literature in various Western, Eastern, and Latin
69
American countries. In a study to identify the economic factors that are responsible for
an increase of crime in Pakistan, Gillani et al. (2009) found that major economic factors,
such as unemployment, inflation, and poverty, are related to the increase of crime in
Pakistan. Gillani et al. also found that the effect of inflation is not only on property
crimes, but also has a positive effect on robbery, which is a violent crime.
Gillani et al. (2009) used a Granger causality test on the assumption that past
events would not be influenced by events occurring in the present or in the future. The
researchers further assumed that if event X happened before event Y, only event X could
cause event Y (Gillani et al., 2009). To determine the effect of inflation on aggregate
crime rates, Seals and Nunley (2007) used data from the FBI's UCR for years 1959 to
2005. While previous researchers based their analyses on aggregate unemployment,
Seals and Nunley based their analysis on inflation. The researchers hypothesized that in
an economic environment with unstable prices, people have an added incentive to use
illegal ways to acquire goods (Seals & Nunley, 2007). Seals and Nunley contended that
since rising prices devalued money, property crime would be an attractive way of making
up for what was lost through inflation. In addition, most of the increase in crime between
1960 and 1970 was largely a result of inflation (Seals & Nunley, 2007).
Unemployment Compensation
Rossie, Berk, and Lenihan's (1980) early study examined the Transitional Aid
Research Project (TARP), based on the idea that crime offenders are sometimes released
but do not have adequate sources to be able to adjust in reentering the society. To
70
compensate for this, TARP put forward a limited amount of income for a limited period.
Rossie et al. found that while the unemployment compensation can reduce recidivism
rates, it can affect one's incentive to work negatively. The payments therefore do not
increase recidivism rates, but the payments can perpetuate unemployment. In fact, the
researchers claimed that the TARP experiment policy can lead to significant support to an
income-maintenance strategy to reduce arrest recidivism among released prisoners.
Summary and Conclusion
The literature review helped determine whether increase in police or police
patrols reduce crime or violent crime, and showed mixed outcomes. The Kansas City
Preventive Patrol Experiment (1974) found that police patrols had no effect on crime and
had no effect on making citizens feel safer. The Newark, New Jersey Foot Patrol
Experiment (1981) found that police patrols had little effect on crime rates. The
experiment, however, found that citizens felt safer in areas with foot patrols. The Flint
Foot Patrol Program (1981) isolated an effect of police patrol on crime and found that
crime levels reduced by 9% because of police patrols. The researchers of the program
found a significant level of satisfaction among the residents for police services. This
literature review showed a progression in the findings from 1974 to 1981; from police
patrols having no effect on crime to having little effect on crime and to having an effect
by reducing crime by 9%.
The Kansas City Gun Experiment Project (1993) found that police patrols had an
effect on violent crime (gun violence), and the seizure of more guns by police resulted in
71
a 49% decrease of gun violence. The researchers of the experiment estimated a
significant decrease in homicide rates. Directed police patrols had a significant negative
effect on gun violence. Citizens were less fearful of crime in the target area but not in the
comparison area. Studies from 1997 had different findings. The Indianapolis Directed
Patrol Project and Levitt (1997) found that police patrols reduced violent crime
significantly. Levitt, however, found that police patrols had a smaller effect on property
crime
The literature review assisted in establishing the researchers who did not find a
correlation or causal effect between increase in police or police patrols and crime
reduction, because of flaws in their methodological approaches. Prior researchers failed
in finding solutions for the endogeneity problem that led them to assert that positive
correlations existed between police and crime (Kelman, Hong, & Turbitt, 2013; Levitt,
2002a, 2002b; Lin 2009; Marvell & Moody, 1996). Recent researches, such as Vollard
(2005), Draca et al. (2008), Zimring (2012), Vollard and Hamed, (2012), and Gill et al.
(2014), found a causal effect between increases of police or police patrols and crime
reduction. These studies presented a significant negative effect of increased police or
police patrols on crime reduction.
In 2001, Zimring (2012) and Kovaudzic and Sloan (2001) found that police
patrols had a negative effect on violent crime and on robbery, burglary, and larceny, but
that police patrols had little influence on homicide rates. Groff et al. (2013) found that
police patrols had no effect on violent crime but had a large negative effect of 75% on
72
property crime. Gill et al. (2014) isolated police as one of the key factors that brought
about the decline of crime in the 1990's, and that about one-fifth of the overall decline of
crime was because of increases in police or police patrols. Telep and Weisburd (2012)
posited that increased police numbers or police patrols reduced crime significantly. The
Pendleton Crime Reduction Project (2005) found that directed police patrols significantly
reduced crime in the target areas. Draca et al. (2008) asserted that increased police
patrols reduced crime, including violent crime, by 11%. Chen and Keen (2014) stated
that increases in police had a significant negative effect on property crime, murder, and
robbery but had no effect on rape and assault.
After a perusal of the literature regarding the effects of police patrols on violent
crime, I highlighted the studies that did not show a correlation or causal effect between
police or patrol increases and crime reduction was because of flaws in their
methodological approaches. Prior researchers failed in finding solutions for the
endogeneity problem that plagued many researchers studying the subject, asserting that
the relationship between police increase and crime was positively correlated (Chalfin &
McCrary, 2013; Rosenfeld & Fornango, 2014; Vollard & Hamed, 2012). Other recent
researchers (Braga & Weisburd, 2011; Nagin, 2013; Telep & Weisburd, 2012) addressed
the problem of endogeneity and found a causal effect between police or police patrol
increases and crime reduction. The literature review also showed that more recent studies
established that a correlation existed between increased police patrols and a decrease in
crime. A causal effect was also established that increased police patrols caused a
73
decrease in crime. The studies did not only find a negative effect of police on crime, but
also established a significant negative effect on crime. These researchers found a
significant negative effect of increased police on crime. These findings are because of
better methodologies or research designs utilized by researchers.
For the literature review regarding the other factors or variables—home-
ownership, unemployment, years of education, cost of living (inflation), unemployment
compensation—a large agreement exists that these factors are related to crime. For
homeownership, Cassidy, Inglis, Wiysonge, and Matzopoulos (2014) found that increases
in home-ownership had a strong and statistically significant effect not only on property
crime, but on violent crime as well. Some researchers believe that years of education are
related to crime. Fella and Gallipoli (2014) found that the more years of education a
person had, the more averse he or she would be to crime, including violent crime. The
researchers asserted that the gap in percentages of incarceration between Blacks and
White (23% higher for Blacks) was because of levels of education, which were higher for
Whites than Blacks (Fella & Gallipoli, 2014).
I analyzed the independent variables by using correlation and multilinear
regression analysis (police patrols, unemployment, cost of living [inflation],
homeownership, years of education, and unemployment compensation) to determine the
relationship between the independent variables and the dependent variable of violent
crime in terms of the strength and direction of the relationship (Kruger, 2011). From the
examination, analysis, and synthesis of the literature review, I established that in more
74
recent studies, researchers have leaned toward the assumption that more police patrols on
the streets of cities are related to and reduce violent crime. This assumption needed to be
empirically tested. The next chapter details the research method (research design) to test
the assumption by using the increase in police patrol with the other variables (factors),
such as years of education, cost of living (inflation), unemployment, home ownership,
and unemployment compensation, to examine the relationship of each factor with violent
crime. The study followed a nonexperimental, quantitative correlational research design.
The choice of design for the study supported the purpose of this research study, which
was to relate the independent variables (individually and collectively) to the dependent
variable. The techniques or tools of analysis were a correlation and multilinear
regression. By using a correlation analysis, I provided a mathematical measurement of
the strength of the relationship between the six independent variables in the study and the
dependent variable of violent crime. Using a multilinear regression analysis helped in the
identification of the way two or more variables were related (Kruger, 2011).
Empirical data regarding the strength of the relationship, positive or negative,
between the independent variables and the dependent variable of violent crime will be
useful to lawmakers, policy makers, and policy analysts because it will guide them in
their various ventures. Since the use of regression allows for the estimation of the value
of one variable from the value of another, the decision maker can tell how an independent
variable may or may not produce an observed change in a dependent variable (Kruger,
2011), information which is critical when allocating resources to fight crime.

75
Chapter 3: Methodology
Introduction
The purpose of this study was to examine the relationship between six
independent variables (police patrols, unemployment rate, level of education, home
ownership, cost of living (inflation), and unemployment compensation) in an aggregated
composite of seven selected major cities with the dependent variable, the rate of violent
crimes. The chapter includes the following sections: research design and rationale,
methodology, threats to validity (including ethical procedures). The chapter ends with a
summary.
Research Design and Rationale
Through this study, I sought to examine the six identified independent variables
and the nature of their relationship to the dependent variable (rates of violent crimes) as
well. The data analysis involved use of multiple linear regression analysis to determine
which (if any) of the independent variables were statistically significant predictors of the
dependent variable in an aggregated sample based on data from seven cities across the
United States. The independent variables were (a) the rate of police patrols, (b)
unemployment rate, (c) years of education, (d) home ownership, (e) cost of living
(inflation rates), and (f) unemployment compensation.
I used a nonexperimental approach for this study because the purpose was to
determine the nature of the relationship between the independent variables individually
with the dependent variable. In addition, I aimed to ascertain which, if any, of the
76
independent variables were statistically significant predictors of the dependent variable
during the time period 2000–2010, based on aggregate data for seven cities, as described
in Chapter 1. Therefore, no intervention was necessary in examining this relationship
because I sought to gather information regarding the naturalistic environment and past
occurrences. In addition, a nonexperimental correlational design was required, since the
study involved seeking answers to the success and failure to control violent crime rates
through increases in police patrols and other factors.
Data collection occurred from several U.S. cities, namely Boston, MA;
Indianapolis, IN; Kansas City, MO; Los Angeles, CA; Philadelphia, PA; Pittsburgh, PA;
and Washington, DC. I retrieved electronically annual violent crime statistics from the
UCR statistics made available by the FBI’s website. Violent crimes were reported in
aggregate, comprised of murder and non-negligent manslaughter, forcible rape, robbery,
and aggravated assault. The FBI website also lists the number of police officers for each
city. Note that civilian law enforcement employees were not included in this figure. I
retrieved the unemployment rate from the Bureau of Labor Statistics (BLS), and retrieved
the average years of education, home ownership rates, cost of living (inflation), and the
weekly unemployment compensation from the U.S. Census Bureau website. Annual data
from years 2000 to 2010 were retrieved for all the variables studied.
A correlation research design provides the ability to examine the relationship of
variables (Ary, Jacobs, Sorensen, & Walker, 2013). Through this study, I sought to
determine the relationship of the six factors stated with violent crime rates. Determining
77
the possible relationships of these variables with violent crimes was the major objective
of the study, but the possible effect of the other relevant factors were not disregarded or
removed from the correlational analysis. Correlation essentially refers to an interest in
determining whether a statistically significant positive or negative linear relationship
exists between two variables (Chatterjee & Hadi, 2015). The purpose of the study was to
determine whether any statistically significant relationships existed between the
dependent variable and independent variables. A statistical correlation design is
appropriate for analysis regarding relationships among and between variables (Cohen,
Cohen, West, & Aiken, 2013; Salkind, 2006).
There are two types of correlation, positive or negative (Cohen et al., 2013). A
correlation can range from -1.0 (perfect negative) to 0 (no relationship) to +1.0 (perfect
positive). A positive correlation between the variables occurs when one variable
increases and the other variable also increases, whereas a negative correlation results
when one variable increases and the other decreases; the opposite would also hold true.
The quantitative research design was appropriate because of the emphasis on
examining whether variables are related to each other (Pickard, 2012). The research
questions for this research study also required descriptive statistics of the study variables
and an analysis of the direct relationship between two variables, which could not be
conducted in a qualitative analysis. Responses provided for open-ended questions, the
typical response in qualitative studies, have to be coded and themes or trends in the
responses have to be determined. A qualitative design did not apply because the study
78
only involved use of scaled or nominal data. The independent variables of interest were
police patrols, unemployment rate, years of education, home ownership, cost of living
(inflation rates), and unemployment compensation. The dependent variable was violent
crime rates.
In contrast to qualitative research in which variables emerge from the data, all of
the study variables were known in advance and hypothesized directional relationships
existed based upon the extant research. Qualitative research is applicable in small
populations for in-depth exploration (McMillan & Schumacher, 2014). Therefore, it was
not applicable, since this research study involved generating conclusions from a big
sample size (N > 798 violent crimes).
Methodology
Population
The population for this study included cities in the United States that have
experienced high levels of violent crime and have increased police patrols in their
respective cities, which include Washington, DC; Indianapolis, IN; Philadelphia, PA;
Boston, MA; Kansas City, MO; Los Angeles, CA; and Pittsburgh, PA. Power is a
function of effect size and sample size (Richardson, 2011). If power is set at .80, and
effect size at .20, then the desired sample size can be calculated. The computation of the
sample size is necessary to determine the statistical power for the study and to check
whether the sample size is sufficient to provide a reasonable power. This was an
79
important consideration for the findings of the research study to become reliable and
conclusive to all cities.
Sampling and sampling procedures. A researcher must obtain the
recommended number of samples based on the computation in a power analysis. When
calculating the sample size for the study, I took three factors into consideration: (a) the
power of the study, (b) the effect size of the study, and (c) the level of statistical
significance. The effect size is a measurement of the strength in the relationship between
the independent and dependent variables (Richardson, 2011). In most instances, the
effect size is defined as small (r = 0.10), medium (r = 0.30), or large (r = 0.50; Cohen,
1977). The power of the study refers to the rejection of Type II error, which is the
rejection of a false null hypothesis. For the study, power was set at 80% (Keuhl, 2000).
The last factor of importance is the level of significance. The alpha level is the
probability of a Type I error, or the probability of rejecting the null hypothesis given that
the null hypothesis is true. Usually alpha level is set to 0.05 or 95% confidence interval
(Draper & Smith, 2014).
The sample size was calculated using G*Power, a general power analysis
program, or a statistical tool specifically used to determine sample size, wherein the
calculation is based on the use of a power of 0.80, a medium effect size (r = 0.15), and on
a level of significance equal to .05. Using the statistical test of multiple linear regressions
with six predictors of the dependent variable of violent crime rates, the resulting
80
computation concluded that the minimum sample size should be 114, which exceeded the
30 samples needed to assume for normality.
Sampling plan. Event sampling methodology allowed me to study ongoing
events that vary across or within time. Event sampling method (ESM) is a form of
sampling method that enables researchers to evaluate ongoing experiences and events
that differ across and within days in its naturally-occurring settings (Pejovic, Lathia,
Mascolo, & Musolesi, 2015). The type of sampling plan selected was appropriate for
studies that required secondary data. Secondary data are readily available data in
historical records, database, and documents. Because the study included historical data, I
aimed to incorporate this variability to examine and determine whether a relationship
existed between the independent variables and the dependent variable. The study
involved annual data retrieved from years 2000 to 2010 for all of the variables studied.
For each independent variable, the frequency of data collected was at a monthly
rate. The dependent variable, violent crime rate, also was collected at a monthly rate. All
data collected occurred at a uniform monthly rate to allow me to examine each
independent variable in relation with the dependent variable.
Data Collection
The instrument used in the data collection was secondary data obtained from
historical records in database and documents. The study did not involve use of surveys
and questionnaires. Data were collected for seven U.S. cities outlined earlier. I retrieved
annual violent crime statistics from the Uniform Crime Reporting (UCR) statistics, made
81
available through the FBI website. Violent crimes were aggregated from the following
categories: murder and non-negligent manslaughter, forcible rape, robbery, and
aggravated assault. The FBI website also provided the number of police officers for each
city. Note that civilian law enforcement employees were not included in this figure. The
BLS website provided the unemployment rates, while the U.S. Census Bureau website
provided the average years of education, home ownership rates, cost of living (inflation),
and the weekly unemployment compensation information for each city.
Based on a power analysis conducted using G*Power, the minimum number of
observations required for the study to have a power of 0.80 was a 114 sample size for
each of the seven cities. This was the sample size recommended by the creators of
G*Power if the study involves a statistical power of 0.80. Increasing the power of the
study to 0.95 would increase the minimum number of observations to 166. This means
that with a power of 0.80, 114 data points would be necessary while for a power of 0.95,
166 data points would be required (Faul et al., 2007). Given the completeness of the
sample of this study, and since a power of 0.80 has been shown to be sufficient for event
studies, this study had a target power of 0.80. Additionally, 114 samples of violent crime
cases in each of the seven cities were enough to have a more reliable and valid measure
of study variables, since the randomness of the data were captured by the many data sets.
I collected the variables of violent crime rate and the six independent variables
that affect the dependent variable, which represented each city included in the sample.
This was to ensure that the findings represented various scenarios in proving that
82
increasing the number of police patrol and other relevant variables in a city would result
in a decrease in the amount of violent crime or not. The minimum number of 114 violent
crime cases for each city based on the G*Power analysis included all of the observations
for each city starting from year 2000 to 2010. I expected that during these periods of
investigation, no city would be dropped from the data set for not meeting the
requirements of 10 years of data. I collected, collated, downloaded, and imported the raw
data into a computer spreadsheet, Microsoft Excel. Each city received a unique
identification number. Each column represented one of the six independent variables and
dependent variable for each month of the 10 years.
Operationalization of Constructs
Data Analysis Plan
Data were entered into SPSS version 22.0 for Windows for analysis. I conducted
descriptive statistics on the data sample’s characteristics, and frequencies and percentages
on categorical data. I conducted means and standard deviations on continuous data,
including number of violent crimes and the six city characteristics: number of police
patrols, unemployment rates, years of education, home ownership rates, monthly cost of
living (inflation), and unemployment compensation rates. The data analysis involved
examination of archival data between 2000–2010 from seven cities (Indianapolis, IN;
Boston, MA; Philadelphia, PA; Kansas City, MO; Pittsburgh, PA; Los Angeles, CA; and
Washington, DC).
83
Research Questions
The study had six research questions. All research questions aimed to examine
the relationship of six independent variables with the dependent variable, rate of violent
crimes. The intention of this study was to be able to answer the research questions within
the context of deterrence theory and economic theory.
2000–2010.
2000–2010.
84
2000–2010.
2000–2010.
Research Question 3: Based on archival data (2000–2010), what is the
relationship the between the level of education (as measured by the average
number of months a citizen of the city has completed school) and the rate of
violent crimes per 100,000 residents in the seven cities examined in aggregate?

85
Research Question 4: What is the relationship between the rate of home
ownership (as measured by the monthly home ownership rates of each city) and
2000–2010.
2000–2010.
86
Research Question 5: What is the relationship between the cost of living (as
measured by annual inflation rates) and the rate of violent crimes per 100,000
residents in the seven cities examined in aggregate during the period 2000–2010.
Research Question 6: What is the relationship between the rate of unemployment
compensation (as measured by the weekly unemployment compensation) and the
aggregate?
period 2000–2010.
87
period 2000–2010.
Study Variables
This section details the variables used in this study beginning with the dependent
variable and the independent variables. The variables in the study were discrete variables
because they were only represented using whole numbers. The discussion includes a
brief description of the variable. The following sections specify the dependent and
independent variables.
Dependent Variable
I calculated the dependent variable by adding the total number of different
offenses involving the use of force. These included murder, robbery, rape, and
aggravated assault. Annual data from years 2000 to 2010 were retrieved.
Independent Variables
The independent variables represented the number of police patrols during the 10-
year period in terms of (a) foot patrols, (b) automobile patrols, and (c) motorcycle patrols.
Annual data from years 2000 to 2010 was retrieved for the variables. These were
aggregated.
88
Unemployment. I measured this by the monthly unemployment rate of each city.
Annual data from years 2000 to 2010 were retrieved for the variable. I retrieved the data
from the Bureau of Labor Statistics.
Years of education. I measured this by the average number of months a citizen
of the city has completed school. I retrieved the data from the BLS. The data were
exported into a spreadsheet.
Homeownership. I measured this by the monthly home ownership rates of each
city. The data was retrieved from the BLS. The data was exported into a spreadsheet.
Cost of living (inflation rates). I measured this with the monthly cost of living
index in a particular city. The cost of living for each month for a given year was
averaged so that the index reflected the inflation rates of the economy.
Unemployment compensation. I used the weekly unemployment compensation
as the independent variable and year as the controlling variable. The total number of
violent crimes was the dependent variable. This data was retrieved from the BLS website
and exported to a spreadsheet.
Preliminary Analysis
Data were screened for accuracy, missing data, and outliers or extreme cases. I
conducted descriptive statistics and frequency distributions to determine that responses
were within the possible range of values and that the data were not distorted by outliers.
The presence of univariate outliers was tested by the examination of standardized values
for the seven variables of interest: rate of violent crimes (dependent variable) and the six
89
city characteristics (independent variables). I created standardized values for each score
and examined cases for values that fell above 3.29 and below -3.29 (Tabachnick & Fidell,
2012). Cases with missing data were examined for nonrandom patterns. Participants
who did not complete major sections of the survey were excluded.
To address Research Questions 1–6, a multiple linear regression determined if the
six city characteristics effectively predicted the rate of violent crimes. The dependent
variable in this analysis was rate of violent crimes per 100,000 residents in the seven
cities examined in aggregate during 2000–2010. Rate of violent crimes was composed
from the sum of different violent offenses, including murder, robbery, rape, and
aggravated assault; it was treated as a continuous variable. The independent, or predictor,
variables in this analysis were number of police patrols, unemployment rates, years of
education, home ownership rates, monthly cost of living (inflation), and unemployment
compensation rates. Numbers of police patrols were composed from the sum of foot
patrols, automobile patrols, and motorcycle patrols, treated as a continuous variable.
Unemployment rates came from the monthly unemployment rates of each city, treated as
a continuous variable. Years of education came from the average number of months a
citizen of the city has completed school, treated as a continuous variable. Home
ownership rates came from the monthly home ownership rates of each city, treated as a
continuous variable. Monthly cost of living came from the monthly cost of living index
in a particular city where the cost of living for each month for a given year was averaged
so that the index reflected the inflation rates of the economy; monthly cost of living was
90
treated as a continuous variable. Last, unemployment compensation rates came from the
weekly unemployment compensation rates, treated as a continuous variable.
A multiple linear regression is the appropriate parametric analysis when the goal
of the researcher is to assess how much variance can be explained by the influence of a
set of independent variables on a continuous dependent variable. This study involved
standard multiple regression (the enter method) to address the research questions. The
standard method entails entering all independent variables simultaneously into the
regression model. Unless theory sufficiently supports the method of entry, the standard
multiple regression is the appropriate method of entry. I evaluated the predictors by what
they added to the prediction of the dependent variable, which was different from the
predictability afforded by the other predictors in the model (Tabachnick & Fidell, 2012).
The F test assessed whether the set of independent variables (i.e., the six city
characteristics) collectively predicted the dependent variable (i.e., rate of violent crimes).
I reported the R-squared—the multiple correlation coefficient of determination—used to
determine how much variance in the dependent variable can be accounted for or
explained by the set of independent variables. The t test determined the significance of
each predictor and beta coefficients were used to determine the extent of prediction for
each of the six city characteristics. For significant predictors, every one-unit increase in
the predictor, the rate of violent crimes increased or decreased by the number of
unstandardized beta coefficients (Tabachnick & Fidell, 2012). Each of the null
91
hypotheses can be rejected if the individual predictor provides a statistically significant,
unique contribution towards the prediction of rate of violent crimes.
I assessed the assumptions of a multiple regression—normality, linearity,
homoscedasticity, and absence of multicollinearity. Linearity is a statistical term that
refers to the linear relationship between one variable or a set of variables to another
variable or set of variables. Linearity assumes that a straight-line relationship exists
between the x-axis (one variable or set of variables) and the y-axis (one variable or set of
variables) so that as one variable or set of variables increases in value, the other variable
or set of variables increases or decreases in values. Linearity is visually examined in
scatter plots. Homoscedasticity is widely known in statistics and means the degree to
which the variance for the dependent variable's scores spread across the values for the
predictor variables. Values should be spread across all various levels to be considered
homoscedastic. Homoscedasticity is visually examined with residual scatter plots (Yang,
2012).
Normality assumes that the scores follow a normal distribution (symmetrical bell
shape). Linearity assumes that a straight-line relationship exists between the six city
characteristics and rate of violent crimes; homoscedasticity assumes that scores are
normally distributed about the regression line (Yang, 2012). I assessed linearity and
homoscedasticity by examination of scatter plots. Researchers use multicollinearity
when discussing inter-correlations among the predictors in the multiple linear regression.
Multicollinearity refers to the extent to which the predictor variables are inter-correlated.
92
According to Kraha, Turner, Nimon, Zientek, and Henson (2012), multicollinearity is
when one or more predictors are highly correlated with one or more other predictors. The
higher the multicollinearity, the lower the chance of finding statistical significance from
the predictor variables. The absence of multicollinearity assumes that predictor variables
are not too related with each other and was assessed using variance inflation factors.
Variance inflation factor values higher than 10.0 suggest the presence of multicollinearity
(Stevens, 2009). If the aforementioned assumptions are violated, the following
nonparametric analysis is conducted: Spearman rho correlations. These correlations are
also known as the Spearman rank correlation. The Spearman correlations are used when
assessing the association between two variables when at least one of those variables is
ordinal or when the assumptions of the parametric Pearson correlations are not met. The
Spearman correlation ranks data values from 1 to N (number of pairs of values) instead of
using precise values. Like most nonparametric analyses, no assumptions are needed for
this test, and the test is less statistically powerful than a parametric analysis.
Ethical Considerations
A researcher who uses human subjects has an ethical responsibility to protect the
participants. When conducting this study, I adhered to the ethical and moral guidelines
outlined by the Institutional Review Board (IRB). The use of archival data eliminated the
necessity to collect identifiable data. In accordance with IRB and federal guidelines, I
safeguarded all data to maintain confidentiality. The safeguard method for data storage is
a locked file within my residence where the data will be held securely for a period of 5
93
years after the research is complete. Upon the completion of the 5-year period, I will
permanently destroy the data.
Threats to Validity
Key threats to validity corresponded to specifics of the sample that provided a
potential bias to the measured results. Confounding variables may have accounted for, or
altered the strength of the relationships among the variables of interest (Howell, 2010).
Because such variables are infeasible to control for the effect of all potential covariates, I
acknowledged this in the interpretation of the findings. As a result, I took additional
caution when interpreting the results and did not assume that the findings can be perfectly
generalized to the population of interest (Creswell, 2009).
In addition, potential limitations exist within the scope of quantitative research.
While quantitative methodologies allow researchers to statistically examine research
questions and hypotheses, such researchers are unable to fully measure the underlying
experiences and perceptions of the participants. Consequently, I traded the degree of
richness present within a qualitative study for a degree of statistical certainty that these
differences did not occur by pure chance (Pagano, 2009).
Summary
This chapter detailed the research design and provided a review of the research
questions in addition to detailing the population and sampling methodology, the
instrumentation and data collection, and the data analysis methodology used for this
study. I determined that a quantitative event study was the most appropriate research
94
design to investigate the relationship between increased police patrol, unemployment
rates, unemployment compensation, home ownership, cost of living (inflation), years of
education, and violent crime rates. Multiple linear regression was the best method to test
the hypotheses. Chapter 4 presents the results of the data analysis to show whether
increased police patrols, unemployment rates, unemployment compensation, home
ownership, cost of living (inflation), years of education, have a relationship on rates of
violent crime in the selected cities. Chapter 5 presents the findings with respect to the
relevant literature.
95
Chapter 4: Results
The purpose of this quantitative study was to determine the nature of the
relationship between the independent variables (police patrols, unemployment rate, years
of education, home-ownership, cost of living [inflation], unemployment compensation,
and the dependent variable (the rate of violent crime) in seven major U.S. cities. The
chapter first presents demographical data followed by descriptive statistics of continuous
variables. The data analysis involved a multiple linear regression, conducted to assess
the research questions. Spearman correlations were used as an ancillary analysis. I
evaluated the significance for all analyses at the generally accepted level, α = .05.
Description of the Sample
Data collection took about 2 – 3 weeks after approval. Violent crime rates ranged
from 0.10 to 0.15, with M = 0.13 and SD = 0.02. The number of police patrols ranged
from 3637.00 to 3837.00, with M = 3708.00 and SD = 71.92. Unemployment rates
ranged from 0.03 to 0.09, with M = 0.05 and SD = 0.02. Average years of education
ranged from 12.05 to 14.10, with M = 13.17 and SD = 0.66. Home ownership rates
ranged from 0.66 to 0.69, with M = 0.68 and SD = 0.01. Cost of living index ranged from
114.20 to 147.50, with M = 134.05 and SD = 11.66. Average weekly unemployment
compensation ranged from 184.00 to 388.00, with M = 295.18 and SD = 75.32. Table 1
presents means and standard deviations of continuous variables. Figures 2–8 present bar
charts for the six city characteristics by year (2000–2010).

96
Table 1
Descriptive Statistics of Continuous Variables
City Characteristics Min. Max. M SD
Violent crime rates 0.10 0.15 0.13 0.02
Number of police patrols 3637.00 3837.00 3708.00 71.92
Unemployment rates 0.03 0.09 0.05 0.02
Average years of education 12.05 14.10 13.17 0.66
Home ownership rates 0.66 0.69 0.68 0.01
Cost of living index 114.20 147.50 134.05 11.66
Average weekly unemployment

184.00 388.00 295.18 75.32
compensation
97
Figure 2. Bar chart for violent crime rates by year (2000–2010).

98
Figure 3. Bar chart for number of police patrols by year (2000–2010).

99
Figure 4. Bar chart for unemployment rate by year (2000–2010).

100
Figure 5. Bar chart for average years of education by year (2000–2010).

101
Figure 6. Bar chart for home ownership rates by year (2000–2010).

102
Figure 7. Bar chart for cost of living index by year (2000–2010).

103
Figure 8. Bar chart for average weekly unemployment compensation by year (2000–
2010).
Detailed Analysis
To address the research questions, I conducted a multiple linear regression to
determine if the six city characteristics effectively predict the rate of violent crimes. A
multiple linear regression is an appropriate statistical analysis when the goal of the
researcher is to assess the relationship between a group of predictor variables and a
continuous criterion variable (Tabachnick & Fidell, 2012). The predictor variables
corresponded to rate of police patrols, rate of unemployment, level of education, rate of
home ownership, cost of living (inflation), and unemployment compensation. The
criterion variable corresponded to violent crime rate per 100,000 residents. Averages for
104
each of the variables were computed from seven U.S. cities by year (2000–2010). Prior
to analysis, I assessed the assumptions of normality, linearity, homoscedasticity, and
absence of multicollinearity.
Normality Assumption
I checked the assumption of normality by examination of a normal P-P lot. The
assumption was met, as the data approximately followed the normality line (see Figure
9).
Figure 9. Normal P-P scatterplot to assess normality for city characteristics and violent
crime rate.
105
Linearity Assumption
I assessed the assumption of linearity by examination of six scatterplots between
the predictor variables and the criterion variable, violent crime rates per 100,000
residents. The scatterplot between number of police patrols and violent crime rates
indicated an inverse relationship (see Figure 10). The scatterplot between unemployment
rates and violent crime rates indicated an inverse relationship (see Figure 11). The
scatterplot between average years of education and violent crime rates indicated an
inverse relationship (see Figure 12). The scatterplot between home ownership rates and
violent crime rates indicated an inverse relationship (see Figure 13). The scatterplot
between cost of living index and violent crime rates indicated an inverse relationship (see
Figure 14). The scatterplot between average weekly unemployment compensation and
violent crime rates indicated an inverse relationship (see Figure 15).

106
Figure 10. Scatterplot to assess for linearity between number of police patrols and violent
crime rate.
107
Figure 11. Scatterplot to assess for linearity between unemployment rate and violent
crime rate.
108
Figure 12. Scatterplot to assess for linearity between average years of education and
violent crime rate.
109
Figure 13. Scatterplot to assess for linearity between home ownership rate and violent
crime rates.
110
Figure 14. Scatterplot to assess for linearity between cost of living index and violent
crime rate.
111
Figure 15. Scatterplot to assess for linearity between average weekly unemployment
compensation and violent crime rate.
Homoscedasticity Assumption
I assessed the assumption of homoscedasticity with a residuals scatterplot (see
Figure 16). The assumption was met. The data represented a rectangular distribution and
no distinguishable pattern existed in the data.

112
Figure 16. Residuals scatterplot for homoscedasticity for city characteristics predicting
violent crime rates.
Absence of Multicollinearity Assumption
The absence of multicollinearity assumption checks that the prediction variables
are not too closely associated with one another. This assumption was tested via (VIF),
where values larger than 10 indicated the presence of multicollinearity and a violation of
the assumption (Stevens, 2009). All of the variance inflation factors values were higher
than 10 for the six predictor variables; therefore, the assumptions were not met. I still
conducted a multiple linear regression, but also conducted Spearman correlations as an
ancillary analysis.
113
The results of the multiple linear regression were significant, F(6, 10) = 11.49, p
= .017, R2 = .945, suggesting that six city characteristics collectively predict violent crime
rates. The R2 indicates that approximately 95% of the variance in violent crime rates can
be explained by the predictor variables. None of the predictors in the model were
significant
The number of police patrols was not a significant predictor of violent crime rates
(t = 2.45, p = .098). Therefore, the null hypothesis (H01) for Research Question 1 cannot
be rejected. Unemployment rate was not a significant predictor of violent crime rates (t =
-0.53, p = .626). Therefore, the null hypothesis (H02) for Research Question 2 cannot be
rejected. Average years of education were not a significant predictor of violent crime
rates (t = 1.83, p = .142). Therefore, the null hypothesis (H03) for Research Question 3
cannot be rejected. Home ownership rates were not a significant predictor of violent
crime rates (t = 2.38, p = .076). Therefore, the null hypothesis (H04) for Research
Question 4 cannot be rejected. Cost of living index was not a significant predictor of
violent crime rates (t = -1.45, p = .222). Therefore, the null hypothesis (H05) for
Research Question 5 cannot be rejected. Average weekly unemployment compensation
was not a significant predictor of violent crime rates (t = -1.39, p = .237). Therefore, the
null hypothesis (H06) for Research Question 6 cannot be rejected. Table 2 presents the
results of the linear regression.

114
Table 2
Multiple Linear Regression with Six City Characteristics Predicting Violent Crime Rates
Source B SE β t p
Number of police patrols 0.00 0.00 1.47 2.15 .098
Unemployment rate -0.27 0.51 -0.24 -0.53 .626
Average years of education 0.11 0.06 3.83 1.83 .142
Home ownership rates 1.99 0.84 1.12 2.38 .076
Cost of living index -0.00 0.00 -2.18 -1.45 .222
Average weekly unemployment compensation -0.00 0.00 -3.59 -1.39 .237
Note. F(6, 10) = 11.49, p = .017, R2 = 0.95.
Spearman Correlations
Because of the absence of multicollinearity assumption not being met, I
conducted Spearman correlations as an ancillary analysis. A Spearman correlation is an
appropriate statistical analysis when the goal of the researcher is to assess the strength of
the relationship between two variables, when one or both of the variables are measured
on an ordinal scale (Pagano, 2009). Data analysis involved Cohen’s standard (Cohen,
1988) to assess the strength of correlation coefficients, where coefficients between .10
and .29 represented a small association, coefficients between .30 and .49 represented a
medium association, and coefficients above .50 represented a large association. A
significant, large, inverse association existed between number of police patrols and
violent crime rates (rs = -.64, p = .035). A significant, large, inverse association existed
115
between unemployment rates and violent crime rates (rs = -.67, p = .023). A significant,
large, inverse association existed between average years of education and violent crime
rates (rs = -.92, p < .001). A significant association did not exist between home
ownership rates and violent crime rates though (rs = .05, p = .883). A significant, large,
inverse also relationship existed between cost of living index and violent crime rates (rs =
-.89, p < .001). A significant, large, inverse relationship also existed between average
weekly unemployment compensation and violent crime rates (rs = -.88, p < .001). Table
3 presents the results of the Spearman correlations between the six city characteristics
and violent crime rates.

116
Table 3
Spearman Correlations between Six City Characteristics and Violent Crime Rates
Variable of interest Violent crime rates
rs p
Number of police patrols -.64 .035
Unemployment rate -.67 .023
Average years of education -.92 <.001
Home ownership rates .05 .883
Cost of living index -.89 <.001
Average weekly unemployment compensation -.88 <.001
Summary
The purpose of this quantitative study was to determine the nature of the
relationship between the independent variables (police patrols, unemployment rate, years
of education, home-ownership, cost of living [inflation], and unemployment
compensation), and the dependent variable (the rate of violent crime) in seven major U.S.
cities. This chapter presented the findings of the data collection. Demographical data
were presented first, followed by descriptive statistics of continuous variables.
Results of the multiple linear regression indicated that the six city characteristics
were collectively significant in predicting violent crime rates. However, none of the
characteristics were significant predictors on their own. Therefore, the null hypotheses
117
(H01–H06) for Research Questions 1–6 could not be rejected. I conducted Spearman
correlations as an ancillary analysis. Average years of education, cost of living index,
and average weekly unemployment compensation had inverse relationships with violent
crime rates. In Chapter 5, I further discussed these findings and compared them to
existing literature. The chapter outlines how the statistical findings connect back to the
theoretical framework. The next chapter also includes suggestions for future research.
118
Chapter 5: Discussion, Conclusions, and Recommendations
The purpose of this study was to determine the nature of the relationship between
the independent variables—police patrols, unemployment rate, years of education, home-
ownership, cost of living (inflation), and unemployment compensation—and the
dependent variable—the rate of violent crime—in seven major U.S. cities. I tested the
hypothesis that a statistically significant relationship existed between each of these
independent variables and the dependent variable (rate of violent crime) using both
correlational and regression analyses. This study was intended to contribute to the
current state of knowledge regarding the historical (2000–2010) relationship between the
aforementioned independent variables and the dependent variable in the seven selected
cities with high rates of violent crime. The data used were open sourced data retrieved
from FBI databases.
Summary of the Findings
Results of the multiple linear regression indicated that the six city characteristics
were collectively significant in predicting violent crime rates. However, none of the
characteristics were significant predictors on their own. Therefore, the null hypotheses
(H01–H06) for Research Questions 1–6 could not be rejected. Results of the regression
were inconclusive because of violations of the assumptions, absence of multicollinearity
and monotonicity (i.e., linear relationship). To calculate interpretable results, I conducted
a series of Spearman correlations. The Spearman correlation is not conducted under the
119
assumption of monotonicity. Because of this, the results could be interpreted with a
higher degree of validity.
Based on the findings from the Spearman correlations, average years of
education, cost of living index, and average weekly unemployment compensation had
significant, large, inverse relationships with violent crime rates. Home ownership rates
had no significant relationship with violent crime rates. However, the results showed a
significant, large, inverse relationship between five of the six variables, the exception
being home ownership rates and violent crime rates.
Interpretation of Findings
The results of a multiple linear regression were significant, F(6, 10) =11.49, p
=.017, R2 = .945, suggesting that six city characteristics collectively predicted violent
crime rates. The R2 indicated that predictor variables can explain approximately 95% of
the variance in violent crime rates. None of the predictors in the model were significant.
A disparity existed between the findings of this study and what other researchers
in the peer-reviewed literature asserted regarding increased police patrols not being a
predictor of lower violent crime rates (Gault-Sherman, 2012; Kleck & Barnes, 2014;
Wain & Ariel, 2014). The findings of this study indicated that (a) unemployment rate
was not a significant predictor of violent crime rates, (b) average years of education was
not a significant predictor of violent crime rates, (c) home ownership was not a
significant predictor of violent crime rates, (d) cost of living was not a significant
predictor of violent crime rates, and (e) average weekly unemployment compensation
120
was not a significant predictor of violent crime rates. A disparity existed between the
findings of this study and that of Zimmring (2012) who found root causes (i.e., poverty,
racism, etc.) to have no effect on the rate of violent crime. This may be because of
differing samples or locations.
Because of the absence of multicollinearity assumption not being met, I
conducted Spearman correlations as ancillary analyses. This was appropriate because the
goal of the researcher was to assess the strength of the relationship between two
variables, when one or both of the variables are measured on an ordinal scale (Pagano,
2009). Cohen’s (1988) standard assessed the strength of the correlation coefficients,
where coefficients between .10 and .29 represented a small association, coefficients
between .30 and .49 represented a medium association, and coefficients above .50
represented a large association. Using this standard, a significant, large, inverse
association was found between the number of police patrols and violent crime rates (rs = -
.64, p = .35).
A significant, large, inverse association existed between unemployment rates and
violent crime rates (rs = -.67, p < .023), which differs from the findings of Entorf and
Spengler (1998) who found unemployment to have little to no effect on violent crime
rates. The large difference between the findings of Entorf and Spengler and this study
indicate that future work may be needed to fully understand why this occurred. A
significant, large, inverse association existed between average years of education and
violent crime rates (rs = -.92, p < .001). This finding was surprising in that both
121
Kleemans (2012) and Tauchen et al. (1994) found education to have no effect on violent
crime rates. Further research may be needed to explore this disparity and complete the
current body of literature regarding this variable.
A significant association did not exist between home ownership rates and violent
crime rates (rs = .05, p = .883). A significant, large, inverse relationship existed between
cost of living index and violent crime rates (rs = -.89, p < .001). A significant, large,
inverse relationship also existed between average weekly unemployment compensation
and violent crime rates (rs = -.88, p < .001). In spite of the significant, large, inverse
relationship that existed between the five variables and violent crime rates, except for
homeownership rates, causation should not be concluded as a result of the relationship
because extraneous factors may be contributing to this relationship. The current body of
literature did not present the aforementioned variables. As such, more researcher in this
field needs to occur in an attempt to thoroughly understand the effect that home
ownership rates and average weekly unemployment compensation have on the incidence
of violent crime rates.
Limitations of the Study
Although quantitative studies can be useful for statistically analyzing research
questions and hypotheses, such studies do not allow researchers to adequately measure
the depth of experiences inherent within a qualitative study. In addition, extraneous
variables may account for the relationships established in the findings. As such, the
statistical findings were interpreted with caution. I focused on establishing the nature of
122
the relationship between the six independent variables in the study and the dependent
variable of violent crime rates, not on causality.
This limitation of a focus on causality was due to the nature of the independent
variables, which could not be manipulated. These variables included number of police
patrols, unemployment rate, average years of education, home ownership rates, cost of
living (inflation), and average weekly unemployment compensation. Not only would it
have been unethical to manipulate these variables, but they could not have been feasibly
manipulated by one researcher, and so they could not be used in a true experimental study
(i.e., a study of causal relationship). The use of a correlational study, which is not useful
for determining causal relationships, was the only methodology available to the variables
of interest. This limitation will be common to all future studies in this area of research.
When computing averages of the different variables, particular cities appeared to
skew the overall mean. For example, Los Angeles had far more law enforcement
employees than any of the other six cities. However, the findings of the study could be
applicable to other cities in the United States in terms of the correlation between the
independent variables and dependent variable of the study. The finding that none of the
independent variables on their own were predictors of violent crime rates, but collectively
predicted violent crime rates, could also be generalizable for other cities. The use of this
study by policy makers in various cities will bring about social change.
123
Recommendations for Future Studies
Based upon the limitations, I derived several recommendations for future studies
from this study:
• While the purpose of the study was to determine the nature of the relationship
between the independent variables (police patrols, unemployment rates, years of
education, home ownership rates, cost of living [inflation], and unemployment
compensation) and the dependent variable (violent crime rates), a causation study
could be carried out by other researchers using the results of this study.
• Although some researchers asserted that police patrols had nothing to do with the
reduction of violent crime rates, this study demonstrated a significant, large,
inverse relationship between them. Future researchers replicating this study may
offer further insight on why this disparity occurred.
• In addition, I found the same result for the other variables, except for home
ownership rates, which had no significant relationship with violent crime rates. In
this regard, it would be useful for other researchers to conduct an integrative study
to determine which of the independent variables caused a decrease or increase in
violent crime rates.
• Because particular cities appeared to skew the overall mean (i.e., Los Angeles),
future researchers should consider studying more cities across the United States or
different cities altogether.

124
• This study did not address displacement of crime; it may be necessary to study
displacement to determine whether the data on the count of violent crime would
have been different if crime displacement was taken into consideration. Crime in
some cases is displaced when hot spot policing is done in a locality, or when
saturated police patrols are implemented in a particular area. Because of this,
future researchers should also consider the displacement of crime.
Conclusions
Research, if carried out with caution and critical awareness, is the key to solving
many everyday problems. Society may hold assumptions that may not be true, but
individuals may act on the assumptions, which can have a huge cost. The determination
of the drivers of violent crime is necessary because huge expenditures are made without
recourse to empirically researched data for guidance in decision-making. Communities
have also experienced immense suffering because of violent crime, which may be
addressed through previous and current findings on this matter. A break from ideological
assumptions on what will solve problems is necessary if people, as a nation, are serious
about reducing the incidence of violent crimes in cities. If decisions made to fight violent
crime are made from empirical data provided by researchers, social change would have
taken place as a result.

125
References
Akesson, B., Smyth, J. M., Mandell, D. J., Doan, T., Donina, K., & Hoven, C. W. (2012).
Parental involvement with the criminal justice system and the effects on their
children: A collaborative model for researching vulnerable families. Social work
in Public Health, 27(1- 2), 148–164. doi:10.1080/19371918.2012.629898
Andresen, M. A. & Malleson, N. (2014). Police foot patrol and crime displacement a
local analysis. Journal of Contemporary Criminal Justice, 30(2), 186–199.
doi:10.1177/1043986214525076
Ary, D., Jacobs, L., Soressen, C., & Walker, D. (2013). Introduction to research in
education. Belmont, CA: Cengage Learning.
Bayley, D. H. (1994). Policing for the future. New York, New York: Oxford University
Press.
Becker, G. (1968). Crime and punishment: An economic approach. Journal of Political
Economy, 76(2), 169–217. Retrieved from: journals.uchicago.edu.
doi:10.1086/259394
Boman, J. H., Krohn, M. D., Gibson, C. L., & Stogner, J. M. (2012). Investigating
friendship quality: An exploration of self-control and social control theories’
friendship hypotheses. Journal of Youth and Adolescence, 41, 1526–1540.
doi:10.1007/s10964-012-9747-x
Box, S. (1987). Recession, crime, and punishment. London, England: Mcmillan
Braga, A. A., & Weisburd, D. L. (2011). The effects of focused deterrence strategies on
126
crime: A systematic review and meta-analysis of the empirical evidence. Journal
of Crime and Research Delinquency. doi:10.1177/0022427811419368
Braga, A. A., Hureau, D. M., & Papachristos, A. V. (2014). Deterring gang-involved gun
violence: Measuring the impact of Boston’s operation ceasefire on street gang
behavior. Journal of Quantitative Criminology, 30(1), 113–139.
http://dx.doi.org/10.1007/s10940-013-9198-x
Brenner, H. (1978). Review of Fox's "Forecasting crime." Journal of Criminal Law and
Criminology, 70.
Byrne, J., & Marx, G. (2011). Technological innovations in crime prevention and
policing. A review of the research on implementation and impact. Journal of
Police Studies, 20(3), 17–40. Retrieved from
https://www.ncjrs.gov/pdffiles1/nij/238011.pdf
Cassidy, T., Inglis, G., Wiysonge, C., & Matzopoulos, R. (2014). A systematic review of
the effects of poverty de concentration and urban upgrading on youth violence.
Health & Place, 26, 78–87. http://dx.doi.org/10.1016/j.healthplace.2013.12.009
Chabay, R. W. (2013). Self-perception and social-perception processes in tutoring: Subtle
social control strategies of expert tutors. In J. M. Olson & M. P. Zanna (Eds.),
Self-inference processes: The Ontario symposium, 6, 217. Abingdon-on-Thames,
United Kingdom: Psychology Press.
Chalfin, A., & McCrary, J. (2013). The effect of police on crime: New evidence from US
cities, 1960–2010 (No. w18815). National Bureau of Economic Research:

127
Washington, DC.
Chapman, J. I. (1976). An economic model of crime and police. Journal of Research
Crime and Delinquency, 13(1), 48–63.
http://dx.doi.org/10.1177/002242787601300105
Chatterjee, S., & Hadi, A. S. (2015). Regression analysis by example. Hoboken, NJ: John
Wiley & Sons.
Chen, Y., Cheng, J., Liang, C., & Sato, M. (2012). Some factors in deviant behaviors of
elementary school students in Taiwan and Japan. Social Behavior and
Personality: An International Journal, 40(4), 623–639.
http://dx.doi.org/10.2224/sbp.2012.40.4.623
Chen, W., & Keen, M. (2014). Does inequality increase crime? The effect of income
inequality on crime rates in California counties. Retrieved from
http://www.matthewkeen.net/documents/ChenKeenAlt.pdf
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2013). Applied multiple
regression/correlation analysis for the behavioral sciences. London, United
Kingdom: Routledge.
Cooter, R., & Ulen, T. (2000). Law and Economics. Boston, MA: Addison-Wesley
Longman.
Corsaro, N., Melde, C., Hipple, N., Cobbina, J., Bynum, T., & Perez, H. (2012). An
assessment of the comprehensive anti-gang initiative: Final project report
(Document No. 240757). East Lansing, MI: Michigan State University, School of
128
Criminal Justice.
Costanza, S. E., & Helms, R. (2012). Street gangs and aggregate homicides. Homicide
Studies, 16(3), 280–307. doi:10.1177/1088767912449623
Creswell, J. W. (2009). Research design: Qualitative, quantitative, and mixed methods
approaches (3rd ed.). Thousand Oaks, CA: Sage.
Currie, E. (1985). Confronting crime: An American challenge. New York, NY: Pantheon
Books.
Cusick, G. R., Havlicek, J. R., & Courtney, M. E. (2012). Risk for arrest: The role of
social bonds in protecting foster youth making the transition to adulthood.
American Journal of Orthopsychiatry, 82(1), 19–31.
http://dx.doi.org/10.1111/j.1939-0025.2011.01136.x
Draca, M., Machin, S., & Witt, R. (2011). Panic on the streets London: Police, crime and
the July 2005 terror attacks. American Economic Review, 101(5), 2157–2181.
http://dx.doi.org/10.1257/aer.101.5.2157
Draca, M., & Machin, S. (2015). Crime and economic incentives. Annual Review of
Economics, 7, 389–408. http://dx.doi.org/10.1146/annurev-economics-080614-
115808
Draper, N. R., & Smith, H. (2014). Applied regression analysis. Hoboken, NJ: John
Wiley & Sons.
Dubourg, R., Hamed, J., & Thorns, J. (2005). The economic and social costs of crime
against individuals and households 2003/2004. London, United Kingdom: Home

129
Office.
Durlauf, S. N., & Nagin, D. S. (2011). Imprisonment and crime: Can both be reduced?
American Society of Criminology, 10(1), 13–54. doi:10.1111/j.1745-
9133.2010.00680.x
Dymnicki, A. B. (2014). The development of an instrument to measure school readiness
for a prevention program. Journal of Research on Adolescence, 24(2), 383–398.
doi:10.1111/jora.12073
Eck, J. E., & Maguire E. R. (2000). Have changes in policing reduced violent crime?: An
assessment of the evidence. In A. Blumstein & J. Wallman (Eds.), The crime drop
in America. New York, NY: Cambridge University Press.
Entorf, H., & Spengler, H. (1998). Socioeconomic and demographic factors of crime in
Germany: Evidence from panel data of the German States (Discussion paper 98-
16). http://dx.doi.org/10.2139/ssrn.155274c
FBI – Data Quality Guidelines: Retrieved from: https://www.fbi.gov/about-
us/cjis/ucr/data_quality_guidelines
Fagan, J. A. (2010). The contradictions of juvenile crime & punishment. Daedalus,
139(3), 43–62. http://dx.doi.org/10.1162/daed_a_00022
Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses
using G*Power 3.1: Tests for correlation and regression analyses. Behavior
Research Methods, 41, 1149–1160. http://dx.doi.org/10.3758/brm.41.4.1149
Fella, G., & Gallipoli, G. (2014). Education and crime over the life cycle. The Review of
130
Economic Studies, 81, 1484–1517. http://dx.doi.org/10.1093/restud/rdu014
Ferguson, C. J. (2010). Violent crime: Clinical and social implications. Thousand Oaks,
CA: Sage.
Fontaine, J., Markman, J., & Nadeau, C. (2010). Promising practices of the District of
Columbia Metropolitan Department (Document No. 1001491) Washington, DC:
Urban Institute
Gault-Sherman, M. (2012). It’s a two-way street: The bidirectional relationship between
parenting and delinquency. Journal of Youth and Adolescence, 41(2), 121–145.
http://dx.doi.org/10.1007/s10964-011-9656-4
Gill, C., Weisburd, D., Telep, C. W., Vitter, Z., & Bennett, T. (2014). Community-
oriented policing to reduce crime, disorder and fear and increase satisfaction and
legitimacy among citizens: A systematic review. Journal of Experimental
Criminology, 10(4), 399–429. doi:10.1007/s11292-014-9210-y
Gillani, S., Rehman, H., & Gill, A. (2009). Unemployment, Poverty, Inflation, and
Causality Analysis on Pakistan. Pakistan Economic and Social Review, 47(1), 79–
98.
Gould, E., Weinberg, B., & Mustard, D. (2002). Crime rates and local labor market
opportunities in the United States: 1977-1997. Review of Economics and
Statistics, 84(1), 45–61. http://dx.doi.org/10.2139/ssrn.121969
Groff, E. R., Johnson, L., Ratcliffe, J. H., & Wood, J. (2013). Exploring the relationship
between foot and car patrol in violent crime areas. Policing: An International
131
Journal of Police Strategies & Management, 36(1), 119–139.
http://dx.doi.org/10.1108/13639511311302506
Hart, T. C., & Miethe, T. D. (2015). Configural behavior settings of crime event
locations. Journal of Research in Crime and Delinquency, 52(3), 373–402.
doi:10.1177/0022427814566639
Hay, C., Meldrum, R. C., & Piquero, A. R. (2012). Negative cases in the nexus between
self-control, social bonds, and delinquency. Youth Violence and Juvenile Justice,
11(1), 3–25. http://dx.doi.org/10.1177/1541204012447959
Hemenway, D. (2014). Guns, suicide, and homicide: Individual-level versus population-
level studies. Annals of Internal Medicine, 160(2), 134–135.
http://dx.doi.org/10.7326/m13-2657
Hoelzer, G., & Gorman, J. (2011). Incorporating hot-spots policing into your daily patrol
plan. The FBI Law Enforcement Bulletin, 80(11), 10–14. Retrieved from
http://leb.fbi.gov/
Howell, D. C. (2010). Statistical methods for psychology (7th ed.). Belmont, CA:
Wadsworth Cengage Learning.
Jenkins, M. J. (2015). Police support for community problem-solving and broken
windows policing. American Journal of Criminal Justice, 1–16.
http://dx.doi.org/10.1007/s12103-015-9302-x
Jiao, A. (2014). The eastern city gun project: Exploring contextual and operational
variables. Journal of Police and Criminal Psychology, 29(1), 10–21.

132
doi:10.1007/s11896-013-9117-y
Johnson, S. D., Guerette, R. T., & Bowers, K. J. (2012). Crime displacement and
diffusion of benefits. The Oxford Handbook of Crime Prevention, 337.
http://dx.doi.org/10.1093/oxfordhb/9780195398823.013.0017
Karandinos, G., Hart, L. K., Castrillo, F. M., & Bourgois, P. (2014). The moral economy
of violence in the U.S. inner city. Current Anthropology, 55(1), 1–22.
http://dx.doi.org/10.1086/674613
Kelman, S., Hong, S., & Turbitt, I. (2013). Are there managerial practices associated with
the outcomes of an interagency service delivery collaboration? Evidence from
British crime and disorder reduction partnerships. Journal of Public
Administration Research and Theory, 23(3), 609–630.
http://dx.doi.org/10.1093/jopart/mus038
Keuhl, R. O. (2000). Design of experiments: Statistical principles of research design and
analysis. Pacific Grove, CA: Duxbury Press.
Kleck, G., & Barnes, J. C. (2014). Do more police lead to more crime deterrence? Crime
& Delinquency, 60(5), 716–738. doi:10.1177/0011128710382263
Kleemans, E. R. (2012). Organized crime and the visible hand: A theoretical critique on
the economic analysis of organized crime. Criminology & Criminal Justice,
13(5), 615–629. doi:10.1177/1748895812465296
Klein, J., Cornell, D., & Konold, T. (2012). Relationships between bullying, school
climate, and student risk behaviors. School Psychology Quarterly, 27(3), 154–
133
169. http://dx.doi.org/10.1037/a0029350
Klick, J., & Tabarrok, A. (2005). Using terror alert levels to estimate the effect of police
on crime. Journal of Law and Economics, 48(1), 267–279.
http://dx.doi.org/10.1086/426877
Knoke, D., Bohrnstedt, G. W., & Mee, A. P. (2002). Statistics for social data analysis.
Boston, MA: Wadsworth.
Kovandzic, T., & Sloan, J. (2002). Police levels and crime rates revisited: A county-level
analysis from Florida (1980–1998). Journal of Criminal Justice, 30(1), 65–76.
http://dx.doi.org/10.1016/s0047-2352(01)00123-4
Kraha, A., Turner, H., Nimon, K., Zientek, L. R., & Henson, R. K. (2012). Tools to
support interpreting multiple regression in the face of multicollinearity. Frontiers
in Psychology, 3. http://dx.doi.org/10.3389/fpsyg.2012.00044
Kruger, N. A. (2011). The impact of economic fluctuations on crime: A multiscale
analysis. Applied Economics Letters, 18, 179–182.
doi:10.1080/13504850903508267
Kulawczuk, P. (1998). The development of entrepreneurship in rural areas. Retrieved
from www.pdc.usprograms/ed/strategy/pdfs/appendix
Larson, R. C. (1975). What happened to patrol operations in Kansas City preventive
patrol experiment? A review of the Kansas City preventive patrol experiment.
Journal of Criminal Justice, 3(4), 267–297. http://dx.doi.org/10.1016/0047-
2352(75)90034-3
134
Lauritsen, J. L., Rezey, M. L., & Heimer, K. (2014). Violence and economic conditions
in the United States, 1973–2011. Journal of Contemporary Criminal Justice,
30(1), 7–28. doi:10.1177/1043986213509024
Levitt, S. D. (1997). Using electoral cycles in police hiring to estimate the effect of police
on crime. The American Economic Review, 87(3), 270–290. Retrieved from
http://pricetheory.uchicago.edu/levitt/Papers/LevittUsingElectoralCycles1997.pdf
Levitt, S. D. (2001). Alternative strategies for identifying the link between
unemployment and crime. Journal of Quantitative Criminology, 17(4), 377–390.
Retrieved from http://www.ppge.ufrgs.br/direito-economia/disciplinas/10/levitt-
2001.pdf
Lösel, F., Stemmler, M., & Bender, D. (2013). Long-term evaluation of a bimodal
universal prevention program: Effects on antisocial development from
kindergarten to adolescence. Journal of Experimental Criminology, 9(4), 429–
449. http://dx.doi.org/10.1007/s11292-013-9192-1
Machin, S., & Marie, O. (2005). Crime and police resources: The street crime initiative
(CEP Discussion Paper No. 680). Bonn, Germany: Institute for the Study of
Labor. Retrieved from http://ftp.iza.org/dp1853.pdf
Machin, S., Marie, O., & Vujić, S. (2011). The crime reducing effect of education*. The
Economic Journal, 121(552), 463–484. http://dx.doi.org/10.1111/j.1468-
0297.2011.02430.x
McMillan, J. H., & Schumacher, S. (2014). Research in Education: Evidence-based

135
Inquiry. Upper Saddle River, NJ: Pearson.
Marvell, T. B., & Moody, C. E. (1996). Specification problems, police levels and crime
rates. Criminology, 34(4), 609–646. http://dx.doi.org/10.1111/j.1745-
9125.1996.tb01221.x
McCrary, J. (2002). Using electoral cycles in police hiring to estimate the effect of police
on crime: Comment. American Economic Review, 92, 1236–1243.
http://dx.doi.org/10.1257/00028280260344768
McElvain, J. P., Kposowa, A. J., & Gray, B. C. (2012). Testing a crime control model:
Does strategic and directed deployment of police officers lead to lower crime?.
Journal of Criminology, 2013, 1–11. http://dx.doi.org/10.1155/2013/980128
Nagin, D. S. (2013). Deterrence: A review of the evidence by a criminologist for
economists. Annual Review of Economics, 5(1), 83–105.
http://dx.doi.org/10.1146/annurev-economics-072412-131310
NARPAC. (2006). Safety and Justice Reports -- 2006. Retrieved from
http://www.narpac.org/DHSJ06.HTM
NARPAC. (2007). Safety & Justice Reports. Retrieved from
http://www.narpac.org/SJR.HTM
O'Donnell, I. (2011). Crime and punishment in the Republic of Ireland: A country profile.
International Journal of Comparative and Applied Criminal Justice, 35(1), 73–
88. http://dx.doi.org/10.1080/01924036.2011.535694
Pagano, R. R. (2009). Understanding statistics in the behavioral sciences (9th ed.).

136
Belmont, CA: Wadsworth Cengage Learning.
Papachristos, A. V., Meares, T. L., & Fagan, J. (2012). Why do criminals obey the law?:
The influence of legitimacy and social networks on active gun offenders. Journal
of Criminal Law & Criminology, 102(2), 397–440. Retrieved from
http://scholarlycommons.law.northwestern.edu/cgi/viewcontent.cgi?article=7426
&context=jclc
Papps, K. L., & Winklemann, R. (1999). Unemployment and crime in New Zealand: New
evidence for an old question. New Zealand Economic Papers, 34(1), 53–71.
doi:10.1080/00779950009544315
Paternoster, R. (2010). How much do we really know about criminal deterrence? The
Journal of Criminal Law and Criminology, 100(3), 765–824. Retrieved from
http://scholarlycommons.law.northwestern.edu/cgi/viewcontent.cgi?article=7363
&context=jclc
Pauwels, L., Weerman, F., Bruinsma, G., & Bernasco, W. (2011). Perceived sanction
risk, individual propensity and adolescent offending: Assessing key findings from
the deterrence literature in a Dutch sample. European Journal of Criminology,
8(5), 386–400. doi:10.1177/1477370811415762
Peguero, A. A., Popp, A. M., Latimore, T. L., Shekarkhar, Z., & Koo, D. J. (2010). Social
control theory and school misbehavior: Examining the role of race and ethnicity.
Youth Violence and Juvenile Justice, 9(3), 259–275.
doi:10.1177/1541204010389197
137
Pejovic, V., Lathia, N., Mascolo, C., & Musolesi, M. (2015). Mobile-based experience
sampling for behavior research. Cornell University Library. Retrieved from
http://arxiv.org/pdf/1508.03725
Pickard, A. (2012). Research methods in information. London, United Kingdom: Facet.
Pietromonaco, P. R., Uchino, B., & Dunkel Schetter, C. (2013). Close relationship
processes and health: Implications of attachment theory for health and disease.
Health Psychology, 32(5), 499–513. http://dx.doi.org/10.1037/a0029349
Piquero, A. R., Paternoster, R., Pogarsky, G., & Loughran. T. (2011). Elaborating the
individual difference component in deterrence theory. Annual Review of Law and
Social Science, 7(1), 335–360. doi:10.1146/annurev-lawsocsci-102510-105404
Police Foundation. (1981). The Newark Foot Patrol Experiment (NCJ No. 081779).
Washington, DC: National Institute of Justice.
O'Connor, T. (2010). Police operations. MegaLinks in Criminal Justice. Retrieved from
http://www.drtomoconnor.com/4000/4000lect06.htm
Raphael, S., & Winter-Ebmer, R. (2001). Identifying the effect of unemployment on
crime. Journal of Law and Economics, 44(1), 259–283.
http://dx.doi.org/10.1086/320275
Ratcliffe, J. H., & Breen, C. (2011). Crime diffusion and displacement: Measuring the
side effects of police operations. The Professional Geographer, 63(2), 230–243.
http://dx.doi.org/10.1080/00330124.2010.547154
Ratcliffe, J., Groff, E., Wood, J., Taniguchi, T., Johnson, L., McGuire-Taylor, C., Sorg,
138
E., & Haberman, C. (2010). The Philadelphia foot patrol experiment research
brief. Philadelphia, PA: Temple University.
Ren, L., Zhao, J., and Lovrich, N. P. (2008). Liberal versus conservative public policies
on crime: What was the comparative track record during the 1990s? Journal of
Criminal Justice, 36(4), 316–325.
Richardson, J. T. E. (2011). Eta squared and partial eta squared as measures of effect size
in educational research. Educational Research Review, 6(2), 135–147.
http://dx.doi.org/10.1016/j.edurev.2010.12.001
Ritchey, M., & Nicholson-Crotty, S. (2011). Deterrence theory and the implementation of
speed limits in the American states. The Policy Studies Journal, 39(2), 329–346.
http://dx.doi.org/10.1111/j.1541-0072.2011.00410.x
Rizzolli, M., & Stanca, L. (2012). Judicial errors and crime deterrence: Theory and
experimental evidence. Journal of Law and Economics, 55(2), 311–338.
http://dx.doi.org/10.1086/663346
Robinson, G. (2014). Drug treatment and desistance from crime (Master’s thesis,
University of Huddersfield). Retrieved from http://eprints.hud.ac.uk/20349/
Rosenfeld, R. (2014). Crime and inflation in cross-national perspective. Crime and
Justice, 43(1), 341–366. http://dx.doi.org/10.1086/677665
Rosenfeld, R., & Fornango, R. (2008). The impact of economic conditions on robbery
and property crime: The role of consumer sentiment. Criminology, 45(4), 735–
769. http://dx.doi.org/10.1111/j.1745-9125.2007.00096.x
139
Rosenfeld, R., Deckard, M. J., & Blackburn, E. (2014). The effects of directed patrol and
self-initiated enforcement on firearm violence: A randomized controlled study of
hot spot policing. Criminology, 52(3), 428–449. http://dx.doi.org/10.1111/1745-
9125.12043
Risman, B. J. (1980). The Kansas City preventive patrol experiment: A continuing
debate. Evaluation Review, 4, 802–808. doi:10.1177/0193841X8000400605
Rutherford, M. B. (2011). The social value of self-esteem. Society, 48(5), 407–412.
doi:10.1007/s12115-011-9460-5
Salkind, N. J. (2006). Exploring research (6th ed.). Upper Saddle River, NJ: Pearson
Education.
Schachter, R. (2014) Public safety partners: Cooperation between college and local police
is expanding, although dealing with sexual assaults on campus remains an
unsolved problem. University Business, 44–49. Retrieved from
http://go.galegroup.com
Schelling, T. C. (1966). The diplomacy of violence. In Arms and Influence (pp. 1–34).
New Haven, CT: Yale University Press.
Seals, R., & Nunley, J. (2007). The effects of inflation and demographic change on
property crime: A structural time series approach (No. 200701). Murfreesboro,
Tennessee: Middle Tennessee State University.
Small, J., & Lewis, C. (1996). Economic crime in New Zealand: Causation or
coincidence? Auckland, Australia: University of Auckland.

140
Tabachnick, B. G., & Fidell, L. S. (2012). Using multivariate statistics (6th ed.). Boston,
MA: Pearson.
Tabarrok, A. (2004). Police, prisons, and punishment: The empirical evidence on crime
deterrence.
Tamayo, A. M., Chavez, C., & Nabe, N. (2013). Crime and inflation rates in the
Philippines: A co-integration analysis. International Journal of Economics,
Finance, and Management, 2(5), 380–385. Retrieved from
http://www.ejournalofbusiness.org/archive/vol2no5/vol2no5_7.pdf
Telep, C. W., & Weisburd, D. (2012). What is known about the effectiveness of police
practices in reducing crime and disorder? Police Quarterly, 15, 331–357.
http://dx.doi.org/10.1177/1098611112447611.
Tita, G., Riley, K., Ridgeway, G., & Greenwood, P. (2005). Reducing gun violence:
Operation Ceasefire, Los Angeles (NCJ 192378). Washington, DC: Department
of Justice. Retrieved from www.ojp.usdoj.gov
Tonry, M. (2011). Less imprisonment is no doubt a good thing. Criminology & Public
Policy, 10(1), 137–152. http://dx.doi.org/10.1111/j.1745-9133.2010.00692.x
Torruam, J. T., & Abur, C. (2014). The relationship between unemployment, inflation
and crime: An application of cointegration and causality analysis in Nigeria.
Journal of Economics and Sustainable Development, 5(4), 131–137. Retrieved
from http://www.iiste.org/Journals/index.php/JEDS/article/viewFile/11277/11578
Umbach, R., Berryessa, C. M., & Raine, A. (2015). Brain imaging research on
141
psychopathy: Implications for punishment, prediction, and treatment in youth and
adults. Journal of Criminal Justice, 43(4), 295–306.
http://dx.doi.org/10.1016/j.jcrimjus.20
The Disaster Center. (2007). US States Crime 2004 -2005 Crimes per 100,000 and
Ranking. Retrieved from
http://www.disastercenter.com/crime/US_States_Rate_Ranking.html
Vollaard, B. (2005). Police numbers up, crime rates down: The effect of police on crime
in the Netherlands, 1996-2003 (CPB Document No. 75). The Hague, Netherlands:
CPB Netherlands.
Vollaard, B. (2006). Police effectiveness: Measurement and incentives. Santa Monica,
CA: The Rand Corporation.
Vollaard, B., & Hamed, J. (2012). Why police have an effect on violent crime after all
evidence from the British crime survey. Journal of Law and Economics, 55(4),
901–924. http://dx.doi.org/10.1086/666614
Wain, N., & Ariel, B. (2014). Tracking of police patrol. Policing: A Journal of Policy
and Practice, 8(3), 274–283. doi:10.1093/police/pau017
Weisburd, D., & Eck, J. E. (2004). What can police do to reduce crime, disorder, and
fear? The Annals of the Academy of Political and Social Science, 593(1), 42–65.
http://dx.doi.org/10.1177/0002716203262548
Weisburd, D., Hinkle, J. C., Braga, A. A., & Wooditch, A. (2015). Understanding the
mechanisms underlying broken windows policing: The need for evaluation

142
evidence. Journal of Research in Crime and Delinquency, 52(4), 589–608.
http://dx.doi.org/10.1177/0022427815577837
Wellford, C. F., & Lum, C. (2014). A new era for hot spots policing. Journal of
Contemporary Criminal Justice, 30(2), 88–94 doi:10.1177/1043986214525078
Wilson, J. Q. (1975). Thinking about crime. New York City, NY: Basic Books.
Yang, H. (2012). Visual assessment of residual plots in multiple linear regression: A
model-based simulation perspective. Multiple Linear Regression Viewpoints,
38(2), 24–37.
Zimring, F. E. (2012). The city that became safe: New York and the future of crime
control. Oxford, United Kingdom: Oxford University Press.

The Relationship Between Increased Police Patrols and Violent Cri PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Relationship Between Increased Police Patrols and Violent Cri PDF

Uploaded by

Copyright:

Available Formats

Walden University

The Relationship Between Increased Police Patrols

Follow this and additional works at: http://scholarworks.waldenu.edu/dissertations

College of Social and Behavioral Sciences

This is to certify that the doctoral dissertation by

has been found to be complete and satisfactory in all respects,

Dr. Mark Stallo, Committee Member,

Dr. Tanya Settles, University Reviewer,

Chief Academic Officer

United States Cities

Victor Sylvester Kabia

MALS, Georgetown University, 2005

BA Hon., Fourah Bay College, University of Sierra Leone, 1981

Dissertation Submitted in Partial Fulfillment

of the Requirements for the Degree of

Public Policy and Administration

correlated to increased police patrols or to the economic variables of unemployment,

inflation, level of education, unemployment compensation, and homeownership. The

inconclusive results. Spearman’s correlations between each of the independent and

United States Cities

Victor Sylvester Kabia

MALS, Georgetown University, 2005

BA Hon., Fourah Bay College, University of Sierra Leone, 1981

Dissertation Submitted in Partial Fulfillment

of the Requirements for the Degree of

Public Policy and Administration

I dedicate this dissertation with affection to my father, mother, and all my

various ways have brought me to this desired point of my quest.

To my family, Marie, Hannah, Samuel R. Kabia, Victoria, John, Sentho, Sinnah,

Namina, I say “Thank You” for believing in me.

To my YUSA family, Samuel B. Koroma, Ibrahim Kamara, Sallu Fullah, Musa

thank you for your encouragement.

List of Tables .......................................................................................................................v

List of Figures .................................................................................................................... vi

Chapter 1: Introduction ........................................................................................................1

Problem Statement .........................................................................................................2

Purpose of the Study ......................................................................................................3

Research Questions and Hypotheses .............................................................................4

Theoretical Framework for the Study ............................................................................8

Deterrence Theory .................................................................................................. 9

Social Control Theory ........................................................................................... 10

Economic Theory of Crime Causation ................................................................. 12

Nature of the Study ......................................................................................................14

Scope and Delimitations ..............................................................................................16

Significance of the Study .............................................................................................17

Chapter 2: Literature Review .............................................................................................23

Theoretical Foundation ................................................................................................25

Deterrence Theory and Crime ............................................................................... 25

Social Control Theory and Crime ......................................................................... 27

Economic Theory of Crime Causation ................................................................. 28

Literature Review Related to Key Variables and Concepts.........................................30

Ineffectiveness of Police Patrols as Deterrence to Crime ..................................... 30

Research Establishing the Effectiveness of Police Patrols ................................... 35

Limited value of traditional patrol methods: Routine and random ................. 36

Effectiveness of directed police patrols .......................................................... 37

Role of foot patrol in crime reduction............................................................. 42

Police Numbers do Matter in Crime Reduction .................................................... 45

Addressing the Endogeneity Problem ................................................................... 54

Other Factors That Affect the Rates of Violent Crime ................................................66

Cost of Living (Inflation) ...................................................................................... 68

Summary and Conclusion ............................................................................................70

Research Design and Rationale ...................................................................................75

Sampling and sampling procedures ................................................................ 79